Munich Personal RePEc Archive

Comparative performance of foreign

affiliates and domestic firms in the Indian

machinery industry

Keshari, Pradeep Kumar

18 May 2013

Online at https://mpra.ub.uni-muenchen.de/33076/

MPRA Paper No. 33076, posted 20 May 2013 21:12 UTC

Comparative Performance of Foreign Affiliates and Domestic Firms in the Indian Machinery

Industry

Abstract

The objective of this paper was to empirically examine the differences in the relative characteristics,

conducts and performance of two ownership groups of firms, foreign affiliates of MNEs (FAs) and

domestic firms (DFs), in the context of Indian machinery industry (IMI) during the period 2000/01 to

2006/07. For this purpose, we applied three alternative techniques, namely, univariate statistical

method based on Welch's t-test comparing the mean value of a variable between two groups of firms,

the multivariate linear discriminant analysis and dichotomous logit and probit models. The common

and significant findings of the statistical analysis suggest that FAs have the greater technical efficiency,

firm size, export intensity, intensity of import of intermediate goods and intensity of import of

disembodied technology but the lower advertisement and marketing intensity and financial leverage.

These findings also give some indications about the quality of FDI that has come to the IMI during the

aftermath of economic reforms. First, it seems that the superior resources and capabilities of FAs

confer them higher technical efficiency (but not overall performance or the monopoly power) and

export intensity in relation to DFs. Second, as the intensity of import of intermediate goods in FAs is

significantly higher than that of DFs, the former group tends to have fewer linkages with domestic

suppliers of intermediate goods including capital goods, raw material, components and spare parts. In

other words, DFs with their activities in the IMI are providing higher linkages with the indigenous

suppliers. Third, the combined results on higher expenses on import of intermediate goods and import

of foreign technology by FAs and no difference in gross profit margins between FAs and DFs point out

that the FAs are probably engaged in the transfer of profits to the MNE system through intra-firm

trade. This aspect, however, require further research which is beyond the scope of this study. Fourth,

despite the higher import of intermediate goods and disembodied technologies, FAs are not spending

higher amounts on R&D towards adaptation or/and absorption of the imported technology and

indigenization of the imported inputs. As a result, R&D intensities of FAs and DFs are the same. Thus,

we conclude that our empirical analysis supports the proposition that the FAs and DFs differ in terms

of the many aspects of conducts and performance in the IMI.

1

Comparative Performance of Foreign Affiliates and Domestic Firms in The Indian Machinery

Industry

Pradeep Kumar Keshari1

1. Introduction

The issue of divergence in the conducts and performance of Foreign Affiliates of Multinational

Enterprises (FAs) and Domestic Firms (DFs) has considerable importance for the policy and decision

makers in the host developing countries and for the researchers interested in the study of Multinational

Enterprises (MNEs) as a separate field of enquiry. One of the major reasons for the recent interest in

this topic is that the world over governments in their respective countries are undertaking substantial

effort and devoting large amounts of resources in the promotion of inward foreign direct investment

(FDI) through its major vehicle the MNEs. This is based on the belief that FAs are superior to DFs in

terms of their holding of firm-specific assets (FSAs) and several measures of performance such as

efficiency and exports (Kobrin 2005). Therefore, locating more FAs in the host developing economies

may lead to direct benefits on account of increased number of firms with superior FSAs and

performance. Besides, the presence of FAs in the host developing economies may indirectly cause

benefits to DFs through horizontal and vertical linkages and by improving their efficiency level and

export performance through increased competition and knowledge spillovers (Smeet 2008 and Görg

and Greenaway 2004).

In view of the above, it is important to understand the origin, nature and the direction of the

differences between FAs and DFs and the effects foreign ownership of the firms have on a host

developing economy, particularly to its domestic sector. The extant literature suggests that the

differences in the characteristics of FAs and DFs and the impacts of the presence of FAs in terms of

generating linkages and knowledge spillovers are contextual, i.e., country or industry specific [Dunning

(2000), Lall and Narula (2004), Jungnickel (2002) and Bellak (2004a)].

In view of substantially increased attractiveness of India for FDI2 and paucity of firm-level and

industry-specific studies in the Indian context, this study attempts to identify various firms'

1The author gratefully acknowledges the encouragements and valuable comments made by Prof. N. S. Siddharthan, MSE, Chennai, Prof. Sunanda Sen and Prof. Pravin Jha, CESP, JNU, New Delhi in writing earlier draft of this paper. The views expressed in this article are entirely personal and does not belong to the organisation to which the author belongs or the scholars who have given their comments on the paper.

2India has become the second most attractive destination (next to China) among MNEs for FDI in terms of A. T. Kearney's 2007 FDI Confidence Index (Global Business Policy Council 2008). World Investment Prospect Survey for 2009-11 places India at the third position after China and the United States in the list of 15 most favoured FDI locations (UNCTAD 2009, p. 38).

2

characteristics, conducts and performance, which could enable a firm to fall in one of the ownership

categories, FAs or DFs, in an industry. Specifically, the objectives of this study are to empirically

examine: a) the differences in FAs and DFs in terms of the major aspects of characteristics, conducts

and performance of a firm which are captured by the several variables: firm size (SZ), age (AGE),

financial leverage (LEV), advertisement and marketing intensity (AMI), capital intensity (CAPI),

research and development intensity (RDI), intensity of import of disembodied technology (MTI),

intensity of import of intermediate goods used for production (MI), export intensity (XI), technical

efficiency (TE) and gross profit margin (GPM).

b) The determinants of probability of the firms to appear as FAs in terms of the above firm-level

variables, while controlling for the sub-industry level influences so as to know if MNE affiliations

make significant difference between the two ownership groups of firms.

The plan for the rest of the study is as follows. In section-2, we define IMI and briefly discuss

the reasons for selection of IMI for this study. Section-3 reviews the relevant literature and formulates

various hypotheses on individual aspects of discriminating characteristics of DFs and FAs and

probability of a firm to fall in the category of FAs (or DFs). Section-4 identifies major characteristics

of the data, sample and period selected for the study. Section-5 explains the statistical methods and

econometric procedures used for the study. Section-6 analyses, discuses and compares the results

obtained from the use of group mean t-test, Linear Discriminant Analysis (LDA) and the estimation of

binary outcome probabilistic (probit and logit) models. Section-7 presents the summary and

conclusions of the study.

2. Indian Machinery Industry-The Focus of Study

Keeping in view the contextual nature of the benefits of FDI, we selected only one industry that

is the IMI - a medium/high technology industry of an emerging economy- for this study. Selection of

only one industry enabled us to reduce heterogeneity across industries arising out of differing product

profiles, levels of product differentiation, industry specific policies, tax and tariff rates, levels of

backward and forward integration, capital intensity, levels of technological capabilities, export

orientations, etc. Focusing on only one industry also reduces heterogeneity in FDI, including the types

and motives of FDI.

IMI represents manufacture of machinery and equipment n.e.c. that is the division 28 in

National Industrial Classification: All Economic Activities-2008 (NIC-2008). The division-28

comprises two types of machinery producing industries, namely, general-purpose machinery (or group

3

281) and special purpose machinery (or group 282) at three digit level of classification. We thus define

IMI as the amalgamation of these two groups of industries.

The major reasons for the selection of IMI inter alia were the following: a) Being a major part

of the capital goods industry, it can be the important source of innovations and higher value addition. It

generally has higher margins and growth prospects as compared to the mature low-technology

industries, in which intense competition has shrunk margins and lowered growth prospects. Being a

technology and skill intensive industry, it could also generate significant intra-industry and inter-

industry externalities and linkages.

b) IMI is relatively under-studied, especially in terms of micro level impact of FDI on its

performance. Besides, there exists no firm-level study to the best of my knowledge that employs

common sample of panel data for the recent period and uses sophisticated econometric methods for

simultaneous examination of several important aspects of comparative behaviour and performance of

DFs and FAs in the IMI.

c) Machinery industry produces higher value-added products, acts as an important source of

innovation and creates strong forward and backward linkages, therefore, the growth of this industry

was considered important by the Indian policy makers.

d) Along with the adoption of outward oriented growth strategy and economic reform measures

implemented since the year 1991, IMI has been exhibiting certain problems including inadequate

technological capability, lack of international competitiveness, global marketing and customer

orientations, management and operational inefficiencies, higher propensity to import than the domestic

production, etc. (CII 2007, EXIM Bank 2008).

e) IMI has received lower level of FDI compared to the other closely related medium/high-tech

industries (viz. electrical machinery and transport equipment) in the post-reform period.3 As a

consequence, during the period of study, FAs as a group constituted only about 20 per cent in the

aggregate sales of this industry whereas FAs' shares are quite high in the other closely related

industries, for examples, 41 per cent in the automobile and auto ancillaries and 42 per cent in the

electrical machinery.4

3Data on cumulative inflow of FDI in India during August 1991 to July 2007 show that: i) the share of manufacturing sector constituted about 56 per cent of cumulative inflow of FDI of about Rs. 2150 million (or USD 50.4 billion) in the country; ii) within the manufacturing sector electrical and electronic equipments (including computer software) received the highest amount with the share of 32.5 per cent, followed by transport equipment industry with the share of 13.6 per cent, chemicals and fertilizers industry with the share of 8.6 per cent and machinery industry with the share of only 5.1 per cent (GoI, 2008).

4These shares are calculated from the data obtained from PROWESS on mean of net sales of each firm for the maximum 7 years and minimum 2 years period between 2000/01 to 2006/07.

4

f) Since machinery industry is categorised as the medium/ high technology industry, the MNEs

could contribute in this industry in a better way either by setting up Greenfield ventures or by offering

latest technology, management and marketing expertise, international business contacts and market

intelligence.

g) The shifting base of machinery and equipment production from the developed to developing

countries has provided major opportunities of production and exports from technologically advanced

countries of the developing economies like China, India, South Korea, etc. The countries like China

and South Korea respectively share 7 per cent and 4 per cent in the world’s total production of

machinery. However, India's share in world's total production machinery is still insignificant 1.4 per

cent, indicating ample scope for expansion in its market share. (EXIM Bank 2008).

3. The Literature, Hypotheses and Variables

The eclectic theory of FDI suggests that FAs and DFs may differ in terms of their competitive

advantages based on the ownership or access to monopolistic advantages, possession of a bundle of

scarce, unique and sustainable resources and capabilities and competence to identify, evaluate, and

harness resources and capabilities from throughout the world and to integrate them with their existing

resources and capabilities (Dunning 2000). This literature also points out that the competitive

advantages of FAs over DFs are partly generic but partly context specific (Ibid).

More specifically, the recent literature suggests the following factors to be generally important

in creating the overall differences in the characteristics of FAs and DFs. First, FAs generally have

privileged access to two types of superior FSA of MNE. The first category of assets is named as

technology type assets including machinery and equipments and skilled labour who operate them. The

technology-type assets can easily spillover to DFs from FAs and the latter may lose competitive

advantage derived from these assets in a short span of time. However, the FAs can maintain their

competitive advantage based on transaction-type FSA for much longer period or until the DFs also

become multinational in their reach (Lall and Narula 2004).

Second, FAs may be more flexible and aggressive in utilising the FSAs, not being hindered by

the inertia that derives from being integrated into the local system, and associated path dependent

political and social obligations (Wang and Yu 2007).

Third, FAs may specialize in a narrow range of activities and operate at a higher end of industry

requiring better technology, skills and OMPs. Thus, the characteristics of the industry segments may be

important in determining their presence.

5

Fourth, MNEs may have acquired indigenous firms having better FSAs and displaying better

performance (than the average DFs in the industry) in terms of R&D, exports, productivity and

profitability performance (Bellak 2004a). This implies that the FAs’ superiority in their average

performance may partly stem from the superiority of the acquired DFs.

Fifth, to the extent positive gaps exist between developed and developing countries in terms of

corporate culture, level of technology, factor endowments and productivity, these gaps may also reflect

in the conducts and performance of FAs with headquarters in the developed countries and DFs based in

the developing economies (Bellak 2004a).

Sixth, by combining location-specific advantages and working in the institutional set up and

policy environment applicable to a host country, FAs may develop their unique set of advantages by

enhancing and modifying the FSAs received from the MNEs. The institutional perspective of business

strategy emphasizes that the resource endowment of the host economy and its institutional framework

moderate the characteristics of FAs, facilitates the development of their resources and capabilities and

even generate new capabilities and new markets opportunities, especially in the emerging economies

(Meyer et al. 2009). Rugman and Verbeke (2001), for example, argue that export from a particular FA

may arise from affiliate specific regional advantages that are grounded in FSA acquired from both the

parent and location-bound advantages.

During the decades of 1990s and 2000s, there has been a growth in empirical literature on

relative performance of FAs and DFs in the manufacturing sector of developed and developing

economies. These studies have mostly used firm-level data and econometric methods. In the case of

developed countries, there have been a few important studies surveying relatively recent literature on

the subject. We focus here on the findings of two important surveys. The first one, Jungnickel's (2002)

edited volume of studies, compares the behaviour of FAs and DFs in a number of European countries.

Second one, Bellak’s (2004) survey, based on the 54 studies mainly using firm-level data in panel

framework, compares the various aspects of performance chiefly for the industries based in the

developed. The research papers in Jungnickel's (2002) edited volume (e.g. Bellak and Pfaffermayr

2002) predominately address both the theoretical and methodological issues associated with

comparison between FAs and DFs. They also empirically tests the differences between FAs and DFs in

terms of selected indicators, such as productivity, wages and R&D. Jungnickel (2002) arrives at two

major conclusions: First, the real difference in behaviour and performance lies between FAs and uni-

national DFs and not between FAs and multinational DFs. Second, the comparison between FAs and

6

DFs is inherently context specific, and hence there are different finding in different countries,

industries, etc.

Bellak (2004) reports that: a) the superior economic performance of FAs over DFs is observed

in the areas of productivity, technology, wages, skills and growth rates but mixed results in case of

profitability; b) the performance gap, however, disappears when firm and industry characteristics are

controlled; c) the gap is more perceptible between FAs and uni-national DFs than between the FAs and

multinational DFs in a host country. Despite the availability of plenty of studies, we find that the most

of the scholars focus on one or a few aspects of firms' characteristics and performance at a time in a

study. There are of course few studies that have tried to examine the differences between FAs and DFs

in terms of the several aspects of firms' characteristics in a single framework. We now turn towards

these studies.

Two noteworthy Indian studies, Kumar (1991) and Ray and Rahman (2006), have examined the

differences in many aspects of firms' behaviour and performance by using the statistical technique of

linear discriminant analysis (LDA). Kumar (1991) empirically examines the issue of differences in the

relative conducts and performance of FAs and DFs in the 43 Indian industries for the early 1980s. The

finding of this study reveals that the FAs are more vertically integrated; have higher access to fund;

operate at larger scales; employ more skilled personnel; earn higher profit margin; and have product

differentiation advantage over their domestic rivals. Based on these results, Kumar (1991) concludes

that the FAs' possession of significantly higher amount of intangible assets (compared to DFs) enables

them to pursue non-price mode of rivalry including product differentiation strategies for maximising

the revenues from these assets. This study, however, is dated and uses aggregated firm-level data for an

ownership category in an industry. In such types of studies, the use of firm-level (or sometimes plant

level) data is considered appropriate (Bellak and Pfaffermayr 2002).

In a relatively recent firm-level study, Ray and Rahman (2006) evaluate the discriminating

conducts of foreign and local enterprises mainly in terms of innovative activities and in establishing

linkages with the domestic (or foreign sector) sector.5 The study uses a stratified random sample of 338

firms, each one with at least Rs. 40 crore of annual sales turnovers for the year 1997/98, belonging to

the Indian chemical, electronics and transport equipment industries. The findings of this study suggest

5The study measures innovatory activities by R&D intensity, import of foreign disembodied technology, and product differentiations through advertising; captures inter-firm forward linkages by distribution outlays as a ratio of net sales turnover, export intensity; approximates backward horizontal linkages by purchase of finished goods as a percentage of sales and import of finished goods as percentages of net sales; measures backward vertical linkages by value added as a percentage of net sales and import of raw materials expenditure as a percentage of net sales.

7

that: a) FAs spend more on import of disembodied technologies than DFs; b) they however do not

significantly differ in terms of R&D intensity, indicating that FAs do not make efforts to adapt their

technologies to the Indian condition; b) FAs foster backward horizontal linkages with local suppliers of

final goods but make less efforts to develop backward vertical linkages. Although Ray and Rahman's

(2006) study uses firm-level data, it excludes performance aspects and does not include IMI in the

scope of their study. Moreover, they are unable to control industry or sub-industry level influences on

the categorical dependent variable capturing the foreign or domestic ownership, probably due to the

limitation of LDA.

Based on the above discussions, we can formulate the following hypothesis constituting the

core of the analysis.

FAs and DFs differ in terms of certain aspects of their conducts and performance due to the

ownership of or access to firm specific assets and their effective utilisation in the context of IMI.

and/or

The firms with certain characteristics are more likely to appear as FAs than DFs

However, the individual aspects of firm characteristics, conducts and performance between FAs

and DFs are equally important. Therefore, we need to have sub-hypotheses on likely differences on the

individual aspects. Further, we predict that the likelihood of a firm to appear as FAs (or DFs) also

depends on the relative characteristics of various sub-industries of the IMI. Hence, the following sub-

sections summarises the theoretical arguments and the findings of empirical literature pertaining to the

individual aspects of relative conducts and performance of FAs and DFs and accordingly forwards

testable hypothesis for this study.

Capital Structure or Financial Leverage (LEV)

Capital structure determines a firm’s value and refers to the way a company finances its assets

through some combination of debt and owned fund. Therefore, capital structure of a firm is represented

by various measures of financial leverage (e.g. long-term debt to networth or long-term debt to total

assets). Since the seminal work of Modigliani and Miller (1958), alternative theories of capital

structure have been developed which include trade-off theory, the pecking order theory and the agency

theory. Besides, a plethora of empirical research has been done to identify the determinants of

corporate capital structure and financial leverage. These researchers have identified firm size, age,

growth prospects, profitability and retained earnings, volatility in earning, tangibility of assets

8

(proportions of fixed assets in total assets), non-debt tax shield (NDTS)6, asymmetry in information

and agency cost, bankruptcy, business and foreign exchange risks, as the major determinants of

corporate capital structure and financial leverage (Akhtar and Barry 2009).

As the FAs and DFs may differ in respect of some of these determinants of capital structure (or

financial leverages), we expect FAs and DFs to differ in terms of financial leverage. In comparison to

DFs, FAs being part of MNE system are expected to have lower volatility in their earnings and

increased access to international capital market, both of which, in turn, would enable FAs to sustain a

higher level of debt without increasing their default risk (Shapiro 1996; Eiteman et al. 1998, pp. 583-

606).

In contrast to the above, the following arguments suggest financial leverage in FAs to be lower

than that in DFs: First of all, as per the Myers’s (1984) pecking order theory of capital structure, if a

firm is more profitable, it is more likely that it would finance its assets more from the internal sources

(e.g. retained earnings which is part of networth or owned fund of a firm), which is easier, readily

available and more cost effective than the external sources. As FAs are expected to be more profitable

than the DFs, the former may retain lower financial leverage. Secondly, the financial and fiscal

expertise coupled with multinationalisation enables better utilization of taxation regulations across

countries and reduction in tax liabilities in MNEs, implying FAs can have higher NDTS than the DFs

(Singh and Hodder 2000). As the tax benefits of maintaining higher leverage are relatively less

valuable for firms with higher NDTS, the FAs (i.e. firms with higher NDTS) are expected to have

lower financial leverage than DFs. Finally, firms with higher agency costs of debt are expected to have

lower debt levels (Jensen and Meckling 1976; Doukas and Pantzalis 2003). FAs' agency costs are

expected to be higher relative to DFs due to higher auditing costs, language differences, and varying

legal and accounting systems (Burgman 1996). In sum, since the some determinants of capital structure

vary between FAs and DFs, the former may have different capital structure than the latter.

There exist several empirical studies comparing capital structure of MNCs and DFs but all of

them are based on the experience of the developed countries. These studies [e.g. Akhtar and Barry

(2009) for Japan; Lee and Kwok (1988), Burgman (1996), Homaifer et al. (1998), Chen et al. (1997),

Chkir and Cosset (2001) and Doukas and Pantzalis (2003) for USA] report that FAs are less leveraged

than the DFs. Mittoo and Zhang (2008), however, find that the Canadian MNEs have higher financial

leverage relative to their domestic counterparts. A study by Akhtar (2005) did not find significant

6As per the accounting practice, the interest on debt is deducted before arriving at net profit while dividend is deducted after net profit. Since the corporate tax is deducted before arriving at net profit, financing through equity is more costly than debt financing.

9

difference in level of leverage between Australian MNCs and Astralian DFs. Hence, the majority of the

studies report FAs to be less leveraged than DFs.

In view of the above reasoning and findings, we hypothesize in the context of IMI that: i) FAs,

as compared to DFs, will be less financially leveraged, and ii) firms with lower value of financial

leverage shall have the higher probability to appear as the FAs.

Firm's Size (SZ)

The size of a firm is a complex variable and may reflect the influence of several factors,

including the amount of resources owned by a firm. Firm size is an indicator of managerial and

financial resources available in the firm, and to the extent that excess resources are available, a firm

will look for opportunities for expansion (Penrose 1959). Besides capturing amount of resources owned

by a firm, the large size acts as an advantage in attracting bigger clients, gathering and processing of

information, achieving economies of scale and scope in production and marketing, exerting political

pressure and winning favours from the government (Mueller 1986, p.139). As substantial resources and

sunk cost are involved in establishing and operating in a foreign location, FAs are likely to be larger

than DFs. Some studies in East Asian countries have found that FAs tend to be relatively large in

comparison to DFs (see Ramstetter 1999a; Takii and Ramstetter 2003).

We hypothesise that i) the average size of FAs would be greater than that of DFs, and ii) firms

with greater size have probability to appear as FAs.

Firm's Age (AGE)

The oldest and biggest firms in the IMI are a few public sector enterprises set up by the GoI

[e.g. Hindustan Machine Tools (HMT), Bharat Earth Movers Ltd. (BEML), Bharat Heavy Electricals

Ltd. (BHEL), and Bharat Heavy Plates & Vessels (BHPV)]. Yet, a major portion of the industry, being

part of the high priority and high technology sector, has been open to foreign participation with

minority equity holding of up to 40 per cent even before 1991 under the old industrial policy regime;

and at least for 51 per cent foreign equity participation on automatic basis since July 1991 under the

new industrial policy7 (Kapila 2001, Chapter 19). Private including foreign participation in this

industry has been increasing after the year 1991 at the cost of public sector participation.

Hence, we may not find any significant difference in the average age of FAs and DFs. We also

predict that the probability of a firm to appear as FA may not be significantly affected by the age of the

firm.

7The prime movers, boilers, turbines, combustion engines and steam generating plants; agricultural machinery; industrial machineries and machine tools have been the part of high priority sector.

10

Choice of Technique (CAPI)

The choice of technique (or technology) of production used by a firm in an industry is generally

captured by its capital intensity. Theoretically, all the firms belonging to an industry, by reasons of

common technology, are expected to operate with the same level of capital intensity. However, the

capital intensity of FAs may be higher than that of DFs for the following reasons. First, DFs economize

on use of capital (than labour) in developing countries because they generally face higher cost in

raising capital (than FAs) in the external market. The opportunity for accessing capital for DFs based in

a developing country is mostly limited to the domestic market, while FAs have better capabilities and

opportunities to raise capital and spread risk globally. However, capital is normally expensive in the

domestic market of the developing countries. Therefore, DFs have to rely more on expensive capital

being available in the domestic market. Even if DFs of a developing country can access capital from

the international market, they have to normally pay higher rate of interest or service charges than that

paid by the FAs. Due to the better corporate image of MNE system, DFs are sometimes crowded out

by FAs even in their own domestic market. Hence, cost of raising capital domestically or

internationally is generally higher for DFs in relation to FAs. Besides, FAs can also access cheaper

internal sources of credit (e.g. the cash flow of the MNE-network) without paying a risk premium

(Oulton 1998).

Secondly, the MNEs originating in the developed countries have comparative advantage in

producing capital-intensive goods. Therefore, their FAs may have affinity towards more capital

intensive industries or more capital-intensive segments of an industry. MNE critics often allege that

FAs do not adapt their capital-intensive technique of production to the labour abundant conditions of

developing countries (Jenkins 1990). The reasons being that: i) FAs are able to pass on the higher cost

of inappropriate technology to the customers due to their market power, ii) there may not be adequate

demand for the product so as to justify the FA’s investment in adjusting the product to the comparative

advantage of the host developing country (Jenkins 1990).

Earlier empirical researches have focused maximum on the choice of technique aspect of the

comparative behaviour of foreign and local firms in the developing countries. Based on a survey of a

large number of empirical studies pertaining to the developing countries, Jenkins (1990) report mixed

results but find considerable evidence about FAs to be more capital intensive than DFs in the

manufacturing sector of Latin American countries, India, Pakistan, South Korea and Kenya. He

concludes that when local and foreign firms are often in direct competition, producing similar products

at similar scale of output, both ownership groups tend to employ equally capital-intensive techniques.

11

There are not many studies examining the issue of choice of techniques for the period of 1990s and

2000s. However, the studies by Ramstetter (1994, 1999a) for Thailand and other East Asian countries

and Ngoc and Ramstetter (2004) for Vietnam suggests FAs to be relatively more capital intensity than

DFs.

We hypothesise that the capital intensity of FAs would be higher than that of DFs and more

capital-intensive firms will have greater probability to appear as FAs in the IMI.

Research and Development Intensity (RDI)

MNEs are well suited for technological innovations and commercialization of technology

generated by other agencies (e.g. research laboratories, universities, etc.) because they have easier and

larger access to financial resources and firm-specific assets, the ability to tap the global market for

scientific and technical personnel, and to organize R&D and to utilise technological assets worldwide

(Dunning 2000). Overwhelming literature on the internationalization of innovative activities suggests

that the MNEs tend to conduct little R&D outside their home base. The MNE literature (see e.g.

Castellani and Zanfei 2006, Chapter-1 and Dachs et al. 2008) offers the following explanations for the

centralization of the major part of R&D activities at the headquarters of MNEs and for conducting only

a minor part of R&D activities by FAs in the host countries.

First of all, R&D generated products and firm-specific assets, including new products or

processes, are mostly created and tested at the respective headquarter locations of MNEs due to: a) the

person embodied nature of knowledge, b) the high level of uncertainty associated with the

development and testing of new products/processes, c) the strong complementarities between the

knowledge base of MNEs and the technological competency of the home-based innovation system, d)

the economies of scale and scope in knowledge production.

Secondly, FAs have privileged access to the stock of technology and R&D laboratories located

at their respective headquarters. Therefore, FAs need to undertake only asset exploiting kind of R&D

activities, involving minor expenditure for absorbing the technology and adapting intermediate goods

obtained from the MNE systems and for customisation of final products to the peculiarities of local

demand, regulations and standards of the host countries.

Thirdly, since technology is a main source of competitive advantages of MNE system,

centralisation of R&D activities at home location enables maintenance of secrecy and avoidance of

leakages to their competitors. Fourthly, MNEs by centralization of R&D avoid coordination costs and

principal-agent problem, which would result if R&D activities are located in different countries.

12

Finally, there could be a lack of scientific infrastructure and highly skilled manpower, particularly in

the developing countries, required for the R&D activities.

Since the 1990s, however, MNEs have been shifting R&D activities from their respective

headquarters to the locations of their FAs in select developing countries, including India and China due

to a set of push and pul factors (UNCTAD 2005 and Siddharthan 2009). It is also reported that the FAs

are complementing the traditional asset exploiting R&D activities with asset augmenting R&D

activities (Castellani and Zanfei 2006 and Siddharthan 2009). The asset augmenting R&D activities

require decentralization of R&D functions of MNE system for exploiting the technological advantages

(e.g. R&D infrastructure and unique accumulated knowledge and inexpensive and high quality skilled

workers) available in the high-tech laboratories and industries of the host countries (Kuemmerle 1999).

Thus, the asset augmenting strategies require FAs to spend more on R&D in their respective host

countries in addition to what is required for the asset exploiting R&D strategies.

Despite the recent trend in the decentralization of R&D activities a large number of empiric0al

studies, relating to both the developed as well as developing countries, reveal that FAs are not more

R&D intensive than DFs. In most of the OECD countries FAs are characterized by lower R&D

intensities as compared to the DFs (OECD 2005). In a study of five small European countries (Austria,

Denmark, Finland, Norway and Sweden), Dachs et al. (2008) find no difference in R&D intensity of

FAs and DFs. In a study on major developing countries of East Asia and Latin America, Amsden

(2001) found that more the foreign ownership less the depth and breadth of R&D. In the case of Indian

manufacturing sector, overwhelming evidences suggest that the R&D intensity of FAs is not more than

that of DFs [viz. Kumar and Saqib (1996), Ray and Bhaduri (2001), Pradhan (2002b), Kumar and

Agarwal (2005), Ray and Rahman (2006), Ray and Venaik (2008), Kathuria (2008), Rasiah and Kumar

(2008)].

In view of the above arguments and findings, we predict that the R&D intensity of FAs would

not be greater than that of DFs and R&D intensity may not be significantly related to the firms'

probability to appear as FAs in the IMI.

Intensity of imported Disembodied Technology (MTI)

Foreign technological collaborations agreements provide a firm foreign technology in

disembodied form, which may include the right to use patents, drawings and designs, technical

services, etc. on payment of royalty and technical fee to foreign technology suppliers. In many cases,

foreign technologies are transferred with supporting documents and know why. Therefore, these

13

technologies can be assimilated, absorbed and used for production purpose with some amount of in-

house technological efforts on the part of firm.

It is normally expected on the basis of FDI theories that FAs should spend less or minimum

amount on import of technology, since they have access to technologies generated within the MNE

system at free or marginal cost. On the other hand, DFs should have higher propensity to import

disembodied technology than FAs. The reason is that the DFs with their limited resources and expertise

are generally incapable of generating new technologies on their own on account of high investment

and sunk cost involved in R&D and risk of failure and appropriation associated with the development

of new technologies. Therefore, DFs generally prefer to import the foreign technologies often available

with MNE system, even if the latter do not sell the latest technologies and put several conditions and

restrictions on the use and absorption of technology.

A firm in the IMI generally uses complex technologies for manufacturing machineries and

equipments. Technological capability of a firm in this industry is determined by product design and

development capabilities and advanced engineering skills. As the DFs in this industry could not

develop these capabilities through in-house R&D, they have depended extensively on import of capital

goods and disembodied technologies for building their technological capabilities. On account of

automatic approval of foreign technological collaboration agreements and lifting of restrictions on

terms of payments and conditions for import of technology in the aftermath of reforms, the firms in IMI

have been heavily depending on import of disembodied technology via foreign technological

collaboration agreements (refer to Chapter-3 of this thesis). In fact, IMI received highest number of

approvals for the foreign technological collaboration agreements during August 1991- July 2007 (Ibid).

The empirical literature on transfer of technology in developing countries suggests that FAs

tend to spend more on import of disembodied technology than DFs [Ray and Rahman (2006), Kumari

(2007), Ray and Venaik (2001 & 2008)]. This may happen for the following reasons: a) MNEs may

transfer technologies free of cost only to those subsidiaries in which they have controlling or hundred

per cent stake and they may supply technology to other affiliates at higher price; b) the sales of

disembodied technologies may boost profit of MNE system as intra-firm trade at transfer prices offers

good opportunity to inflate price of the technology and intermediate goods to be supplied to the foreign

affiliates; c) MNE system normally allows FAs to undertake only minor kinds of R&D activities,

therefore, they obtain free or/and purchase disembodied technologies developed at the headquarters of

their respective MNE systems.

14

It is thus hypothesized that FAs spend greater amount on import disembodied technologies than

that of DFs and firms with higher intensity to import disembodied technology will have greater

probability to appear as FAs in IMI.

Advertising and Marketing Intensity (AMI)

Advertising and marketing tactics and R&D are considered as the two major elements of non-

price strategies followed by MNEs for differentiating their products and competing with their rivals.

FAs are expected to follow more intensive advertising and marketing strategies to promote sales of

their products than what is followed by DFs. Against this logic, one may also expect FAs to be

pursuing less intensive advertising and marketing strategies than those adopted by DFs in the IMI for

the following reasons: i) In the international as well as Indian market, brand equity of products sold by

FAs and corporate image of MNE system may have already been established and thereby MNE system

to which FAs belong may be well known as a reputed supplier of producer goods. Therefore, it may

not be necessary for FAs to spend substantial amount on current advertising and marketing; ii) FAs

may be concentrated in segments of machinery industry, which may not require substantial advertising

and marketing campaign for the enhancement of sales. Instead, these segments may depend more on

increased efforts towards R&D for product differentiation and adaptation.

Only a small number of empirical studies have compared advertising intensity of FAs and DFs

in the industrial sector and findings of these studies are not conclusive (Jenkins 1990; Kumar and

Siddharthan 1997). However, most of these studies do not control for other firm or industry-specific

characteristics while comparing the advertising intensity of DFs and FAs (Jenkins 1990). Advertising

and marketing, a phenomenon associated with imperfectly competitive market, are used as a means to

reduce the scope and effectiveness of price competition by creating product differentiation and strong

goodwill for the firm. Advertising and marketing are widely accepted as the most effective methods of

product differentiation among firms in consumer goods industry. In a producer goods industry like

machinery industry, advertising and marketing expenses related to selling of goods may be less

important in creating product differentiation. We, therefore, capture product differentiation advantage

of a firm in machinery industry partly by its advertising and marketing intensity (AMI) and rest by

R&D intensity.

In view of the above arguments, we feel that the AMI of DFs and FAs may not differ

significantly in the IMI.

Export Intensity (XI)

15

FAs have the following advantages over DFs in undertaking exports (Greenaway and Kneller

2007, Kneller and Pisu 2007): First, FAs’ access to superior technology and organisational and

management practices leads to higher productivity8, cost competitiveness, better quality and quick

delivery of their products and after sale services. Secondly, production and marketing network of the

MNE system itself provides an outlet for the intermediate and final products of FAs. Thirdly, entry in

third country export market requires incurring sunk cost. Since MNEs are better placed than DFs in

terms of financial resources and have already incurred major part of sunk cost by virtue of

multinational scope of their operation, FAs may find it easier (than DFs) to penetrate in the

international market, particularly in the markets with high barriers to entry or of highly differentiated

and technologically sophisticated products. Fourthly, FAs are better equipped to resist protectionist

pressures in their home countries in such a way as to favour imports from their affiliates (Helleiner

1988).

Against the above arguments, there are the following reasons to believe that the export intensity

of FAs may not be more than that of DFs. First of all, a MNE operates with the help of its worldwide

network so as to maximise the global profits but not necessarily the profits of its individual subsidiaries

(Hymer 1976). Thus, a parent MNE, which has control over its FAs, may not allow them individually

to maximise exports and profits resulting from exports, if these are expected to reduce the MNE's

global profitability. This is sometimes accomplished by under pricing the exports from MNE affiliates

to parent firm or to other affiliates in the MNE’s network.

Secondly, technology transfer and financial agreements between the MNEs and their FAs often

include restrictive clauses controlling the export behaviour of the latter. A RBI (1985) study on Indian

manufacturing sector has pointed out high incident of restrictive clauses either totally prohibiting or

strongly limiting the latter's exports. Thirdly, if the nature of FDI is market seeking, export intensity of

FAs and DFs may not differ significantly (Nayyar 1978). Fourthly, if FAs suffer from higher cost of

production relative to their parents and others affiliates in the parents network, their ability to export

would be limited (Abdel-Malek 1974). This may happen if FAs are greatly affected by liability of

foreignness and FAs are unable to develop FAs specific advantage.

Most of the older empirical studies examining the export performance of FAs vis á vis DFs in

the manufacturing sector of developing countries have shown mixed results and used divergent and

unsatisfactory methodologies of comparing FAs and DFs (refer to Casson and Pearce 1987 and Jenkins

1990 the surveys of literature). The recent studies on developing countries which mostly use firm-level

8Finding in this indeed shows that FAs are more technically efficient than DFs.

16

data and econometric techniques indicate FAs to be more export oriented than DFs. These studies

include Ramstetter (1999a and 1999b) on selected East and South East Asian Countries; Sun (2009),

Du and Girma (2007) and Fung et al. (2008) for Chinese manufacturing; Lutz and Talavera (2004) on

Ukraine; Jensen (2002) on Poland; Rasiah (2005) for textiles and garments, food and beverages and

metal engineering firms in Kenya; Rasiah (2004) for electronics exporting firms in Malaysia,

Phillipines and Thailand; Chudnovsky and Lopez (2004) for MERCOSUR countries; Ngoc and

Ramstetter (2004) for Vietnam; Rasiah and Malakolunthu (2009) for electronics exporting firms in

Malaysia; Wignaraja (2008a) for a sample of clothing firms in Sri Lanka; Correa et al. (2007) for

Ecuador. Kumar's (2005) literature survey on Indian studies reveals statistically insignificant difference

in the export performance of FAs and DFs during pre-reform period in majority of the cases.9

Indian studies pertaining to post-reform period report mixed results. Employing a cross-section

spline regression method, Chhibber and Majumdar (2005) concludes when property rights devolves

unequivocally to foreign owners (i.e. with majority ownership of equity) the Indian firms display

higher export orientation. In the case of Indian information technology sector, Siddharthan and Nollen

(2004) report that the export intensity of FAs is greater than that of DFs. Bhaduri and Ray's (2004)

firm-level study provides weak evidence on FAs to be more successful in exporting than the DFs in

Indian pharmaceutical industry but find no difference in export intensity of FAs and DFs in the case of

electrical/electronic industry. Using OLS method, Rasiah and Kumar (2008) report FAs to be better

than DFs in terms of export intensity in automotive parts industry. Ray and Rahman (2006) and Ray

and Venaik (2008), however, came to the conclusion that FAs are less export intensive than the DFs

belonging to the chemicals, electronics and transport equipment industries.

In view of the above discussions, we hypothesize that: i) FAs are more export intensive than

DFs, and ii) more export intensive firms shall have greater probability to appear as FAs.

Intensity of imported Intermediate Goods (MI)

A firm can procure capital goods, raw material, components and spare parts locally or import

the same. In the former case, the firm creates backward linkages, which helps in building additional

capacities for production of raw material, components and other intermediate inputs, etc. in the host

economy. It is said that the FAs have fewer linkages with the host economy than the DFs as the former

on the average maintains the higher intensity of imported intermediate goods (MI) than the latter based

in the same industry. Based on the literature on FDI and import of MNEs, the following explanations

for higher import orientation of FAs over DFs can be offered.

9Refer to Appendix-1 of Chapter-7 for details about the studies conducted during the pre-reform period.

17

First of all, FAs normally perceive the reliability and quality of supply in the host developing

country to be inferior. Therefore, they prefer to source their input requirements including machinery,

raw material and components from the MNEs system including parent and its affiliates (Rugman 1981,

Hennart 1986). The use of imported and superior raw materials and capital equipments ensures better

quality of products leading to barriers to entry (or mobility) through differentiation advantage (Ray and

Venaik 2001). The importance of quality factors may be more important in the context of machinery

industry, since the efficiency of the user industries of machinery industry largely depends on the

quality, reliability, durability, precision and overall efficiency of machineries and equipments supplied

by the machinery industry.

Secondly, even if cost, quality and reliability of supplies are the same, a MNE affiliate may

prefer to obtain inputs from its parent or parent's network so that the parent can capture supplier's

profits and utilize economies of scale in production and distribution. Besides, continuing to import

intermediate inputs provides opportunities for transfer pricing which may be lost with local sourcing

(Jenkins 1990).

Thirdly, it is possible that the MNE may have preferential access to relevant raw material and

machinery used for the production and may be operating in a product segment, which has less vertical

linkages in the host's market. Finally, MNEs may have interest in maintaining high import content to

please home country trade unions and the governments respectively worried about jobs and about trade

deficits or loss of production and employment to a foreign country (Natke and Newfarmer 1985; Natke

1987).

In the high technology industry such as machinery, DFs may also depend on imports for

sophisticated machinery, capital goods and other critical inputs but they are less likely to be tied to the

overseas supplier. They will attempt to indigenise the imported items as soon as possible so that they

can capture the suppliers' profits. In case the inputs are available in the local market, DFs may procure

the inputs from the local producers to ensure the timely supply rather than bothering too much about

quality of the supply. Further, the DFs may not prefer importing because they may not be well equipped

to bear or tackle the uncertainty of exchange rate fluctuations and hassles of importing from the

international market about which they obviously have less information than a MNE. Furthermore, DFs

normally operate on the lower end of the industry that may not require such sophisticated technologies,

capital goods and raw materials, etc. for which they have to depend heavily on import.

The majority of the earlier studies in developing countries reveal that the FAs are more import

intensive than DFs (Jenkins 1990 and Siddharthan and Kumar 1997). The latest studies on Indian

18

manufacturing sector and the literature survey therein [viz. Ray and Venaik (2001) and Ray and

Rahman (2006)] report that FAs are more import intensive than DFs. In view of the above arguments

and the findings of the empirical literature, we hypothesize that FAs may have higher MI than DFs and

firms with more import orientation have probability to appear as FAs in the IMI.

Technical Efficiency (TE)

Bellak and Pfaffermayr (2002), Bellak (2004) and others identify the following major reasons

for higher productivity/efficiency performance of FAs as compared to DFs: First and foremost, FAs,

being part of MNE system, have access to firm-specific assets10 (e.g. newer and superior technology,

organisational and management practices) at marginal cost and to the internal market of the MNE

systems. Therefore, FAs benefit from the productivity/efficiency spillovers of the system and multi-

plant economies of scale. FAs may also develop their unique sets of productivity enhancing FSAs

while applying the FSAs accessed from their respective MNE systems to the locational conditions of

the host countries.

Second, FAs specialize in narrow spectrum of activities due to strategy of MNEs to fragment

the production stages internationally according to the locational advantages of the host countries. FAs

normally exist in higher end of an industry requiring intensive use of superior FSA, whereas DFs may

exist in lower end of production involving standard technology and lower skill levels. For instances, on

account of the availability of cheaper skilled workers in India, FAs may undertake highly technical or

core activities with automated production facilities in a sub-industry of machinery industry requiring

highly trained staff with above average efficiency. As most of the DFs in our sample do not have

transnational presence, they are unable to fragment the production stages internationally.

Third, DFs may select and adopt inferior technology while FAs may use frontier technology.

For example, import of second hand machinery has increased substantially in India after its

liberalisation (refer to Chapter-3 of this thesis). Compared to FAs, DFs may have higher propensity to

use inferior machineries for the lack of adequate information about the frontier technology and lack of

financial resources needed for acquiring the frontier technology, price sensitivity of their customer,

inadequate market size or clientele for the quality products and unavailability of best practice

technology due to strategies of the MNEs.

Fourth, MNEs would have formed FAs by acquiring more productive plants or firms possessing

unique strategic FSA in IMI. Therefore, FAs may enjoy higher productivity than DFs.

10Expenses on generation and development of FSA and auxiliary services like training, controlling, etc are counted as expenses of the headquarter but the FAs derives the benefits of the same without incurring any cost or by incurring minimal cost. DFs, although they may operate affiliates, have to bear the full cost of such assets or services.

19

Fifth, MNEs follow superior corporate governance practices as compared to DFs. Therefore,

the top managements in FAs may be under higher pressure to perform and show better efficiency than

the management of DFs, especially after MNE’s takeover of a local firm through a strategic

investment.

Sixth, FAs have access to financial capital of MNE system, which makes financing of the

business of FAs easier and cheaper compared to that of DFs.

Seventh, as compared to DFs, FAs generally employ and retain highly skilled workers by

paying them higher wages and by constantly upgrading their skills through regular trainings and

exposure to best-practices in the industry.

Eighth, since the MNEs have global outlook, they are able to respond quickly to the changes in

the policy environment, emerging opportunities and locational advantages of a country. For instances,

they may invest and divest plants frequently, achieve better match between locational advantages and

FSAs, cherry pick plants/firms with above average productivity in an industry. This is almost

impossible by uni national DFs and possible to a much lesser extent by newer MNEs headquartered in

a developing country.

Ninth, the gap in the productivity/efficiency between the home country of a FA and the host

country may be reflected in the gap in productivity/efficiency of FAs and DFs. Thus, the TE of FAs

may also be higher than DFs because FAs are linked to MNEs headquartered in the developed home

country and DFs are based in a developing host country like India. It may be noted that the average

labour productivity of Indian manufacturing firms are lower compared to the other countries of

emerging market economies (Lakshmanan, et al. 2007).

Several studies pertaining to the period 1990s and 2000s for the developing countries report

FAs to be more productive than DFs [e.g. Blomström and Wolff (1994) for Mexico; Okamoto and

Sjöholm (1999) and Sjöholm (1999a), Takii (2004), Takii and Ramstetter (2003) for Indonesia; Haddad

and Harrison (1993) for Morocco; Kokko et al. (2001) for Uruguay; Ramstetter (1999a) for East Asian

countries; Chuang and Lin (1999) for Taiwan; Hallward-Driemeier et al. (2002) for various East Asian

Countries11; Ngoc and Ramstetter (2004) for Vietnam; Sinha (1993), Kathuria (2001), Ray (2004),

Goldar et al. (2004), Sasidharan and Ramnathan (2007) for Indian manufacturing sector].

On the contrary, some studies [e.g. Patibandala and Sanyal (2005) for Indian manufacturing

sector; Ito (2002), Ramstetter (1994, 2002b, 2003), Tambunlertchai and Ramstetter (1991) for

11Hallward-Driemeier et al (2002) used questionnaire survey covering 2700 manufacturing firms from the five East Asian countries Indonesia, Korea, Malaysia, the Philippines and Thailand. Their regression analysis revealed that, even after controlling for sector, size and export orientation, FAs have higher productivity than DFs in all countries except Korea.

20

Thailand; Menon (1998) and Oguchi (2002) for Malaysia; Konings (2001) for Bulgaria and Rumania]

suggest that FA are not more productive than DFs.

In this study, we capture the efficiency by a measure technical efficiency which for a given firm

(in a given year) is defined as the ratio of its mean output (conditional on its level of factor inputs and

firm effects) to the corresponding mean output if the firm utilizes its levels of inputs most efficiently

(Battese and Coelli1992). This measure of technical efficiency by design has values between zero and

one. The method and derivation of technical efficiency is explained in Appendix-1.

Gross Profit Margin (GPM)

The reasons for higher profitability in case of FAs compared to DFs may be the following

(Jenkins 1989). First of all, as discussed in the last sub-section, FAs may enjoy higher technical

efficiency/productivity. Secondly, FAs may face favourable demand conditions for their products in

developing countries whether they enter into an existing industry (either through Greenfield venture or

acquisition) or an entirely new industry. In the case of an existing industry, FAs may set price initially

in line with the higher average costs generally prevailing in the industry. Since FAs have cost

advantage over existing DFs, the former enjoy surplus profits. In the case of a new industry, where the

demand conditions are quite favourable in relation to supply, FAs would be able to charge a high price

and thereby earn higher rate of profit.

Thirdly, customers of developing countries may also perceive products of MNEs as superior in

terms of non-price attributes such as quality, technological sophistication, reliability, durability, just-in-

time delivery and after-sales service even if they may not mind paying higher than market price for the

same. Finally, as explained by Kumar (1990), FAs and DFs constitute two different strategic groups in

Indian manufacturing sector. Further, the group of FAs enjoys greater protection from “mobility

barriers”12 and thereby attain greater profitability on account of market power, notably in the

knowledge-based industries.

Empirical evidence concerning the existence of profitability differential between DFs and FAs

is mixed but in majority of the cases FAs outperform the DFs in terms of profit performance. Jenkins

(1989) in his survey concluded that FAs do enjoy higher profitability (than the DFs) based in the

manufacturing sector of the developing countries, mainly on account of their productivity advantages

and higher demand for their products. However, these studies are quite dated and use rudimentary

methods of comparisons. Bellak's (2004) survey includes more recent studies which employ

12Mobility barriers are defined as entry barriers, which not only impede fresh entry to the industry but also restrict inter strategic group mobility of the existing firms. Thus, firms in a particular strategic group may not only enjoy protection from new entrants to the industry but also from existing firms belonging to other strategic groups in the same industry (Kumar 1990).

21

econometric methods for comparing the profit performance of FAs and DFs. However, he too finds

mixed results. He explains the reasons for mixed results in terms of differences in the quality of data

used across the studies and rent shifting through the use of transfer pricing mechanism adopted by the

MNEs.

Some studies in the context of East Asian countries [e.g. Wiwattanakantang (2001) for

Thailand, Ramstetter (1999a), Ramstetter and Matsuoka (2001) for other ASEAN countries] suggest

that FAs enjoy higher profitability than DFs. Similarly, Anastassopoulos (2004) in the case of Greek

food industry finds that the profitability of FAs to be higher than that of DFs even after controlling for

other determinants of profitability. A recent study by Aydin et al. (2007) on all the quoted firms on

Istanbul Stock Exchange, Turkey and literature survey therein reveals that the FAs perform better than

local firms. In contrast, a study by Barbosa and Louri (2005), employing a quantile regression analysis

suggests that foreign ownership ties in general do not make a significant difference with respect to

performance of firms operating in Portugal and Greece.

In the context of Indian manufacturing sector, several authors including Kumar (1990),

Chhibber and Majumdar (1999) and Douma et al. (2006) reveal significant association between foreign

ownership and firm’s performance, measured by various indicators of profitability including gross

profit margins. Based on the regression analysis of industry level data covering 43 Indian

manufacturing industries, Kumar (1990) found that the profits before taxes as a proportion of sales was

higher for FAs than for DFs even after controlling for other influences on profitability. He explained

the reason for superior profitability of FAs in terms of greater protection enjoyed by FAs from ‘entry

and mobility barriers’ leading to greater market power rather than the higher ability (or efficiency) of

FAs.

Chhibber and Majumdar (1999) show that, after controlling for a variety of firm and

environment-specific factors, only when property rights devolve to foreign owners, at ownership levels

providing unambiguous control at 51 percent, foreign owned firms display relatively superior

performance as compared to domestic firms in the Indian corporate sector. Controlling for firm size,

age, business group affiliation and industry specific effect, Douma et al. (2006) tested the impact of

foreign ownership on performance of 1005 Indian manufacturing firms in 1999 and 2000 by applying

OLS multiple regression method. They observed that foreign ownership positively affects the firms

profitability measured by return on assets.

Based on the above arguments and on the review of empirical literature, we put forward the

following hypotheses. First, FAs may show greater performance than DFs in terms of two important

22

firm-level performance variables, namely, TE and GPM. Second, firms with higher TE and GPM may

have greater probability to appear in the group of FAs than in the group of DFs.

Index of Market Concentration (IMC)

Hymer (1976) stresses that the MNEs are prevalent in concentrated markets where the few

firms command major share of the sales (Caves 1996, chapter 4). In such markets, sellers are not price

takers; and the best response of each seller is conditional upon the actions of other sellers. Lall (1978 &

1979) and others suggest that the operations of FAs are likely to increase the industrial concentration in

the long-run and thereby they may be found mostly in the concentrated industries. The following

factors are considered chiefly responsible for this phenomenon: (i) inefficient small firms may exit or

merge in the face of increased competition from FAs having competitive advantage over DFs; (ii) FAs

may use their privileged access to financial resources to outlast their rival by resorting to price and non-

price warfare, and predatory practices. The distortions in market for firms considerably favour MNEs

in buying out of local companies (Newfarmer 1983); (iii) the conducts of FAs may have an indirect

effect on concentration by stimulating defensive amalgamations among DFs and raising barriers to

entry for new entrants.

The TCI approach of FDI, however, seems to suggest that entry of MNEs creates more

competition and breakdowns the existing oligopolistic structure, particularly in the developing

countries. Therefore, it is more likely that FAs are present in less concentrated and more efficient

industries. Hence, it is difficult to predict whether firms in a concentrated industry or sub-industry will

have more (or less) probability to observe as FAs.

IMI constitutes 24 product groups which differ amongst each other in terms of the level of top

four-firm sellers concentration ratios. A firm in IMI may predominantly operate in one or in a few of

these product groups. Some of these product groups may have more presence of FAs while other may

have less. To examine the relationship between market concentration and probability of the firms to

appear as FAs, we have devised a firm-specific index of market concentration (IMC), which is a

weighted average of the four-firm sellers’ concentration ratios of each of the product groups in which a

sample firm predominantly operates.

Sub-Industry Level Influences

To minimize the industry-level influences on the probability of a firm to appear as FA, we have

selected a single industry, the IMI. Yet, this industry covers several sub-industries, which may differ

among each other in terms of product profile, demand conditions, and the barriers to entry stemming

from the technological sophistication, choice of techniques of production, level of product

23

differentiations, minimum efficient scale of production and initial capital required for setting up plant,

gestation period, etc. The sub-industries may also differ in terms of productivity, profitability and

growth prospects, etc.

MNEs are better placed due to their asset advantages (than DFs) to overcome barriers to entry

in a host country industry. It is also observed that the MNEs are more attracted towards industries with

four characteristics, notably, high levels of R&D relative to sales; large share of professional and

technical workers in their work forces; products that are new and /or technically complex; products

with high levels of differentiations created through advertising, marketing and other means (Markusen

1995, UNCTAD-WIR 2005). Hence, a firm has probability to appear as FAs, it belong to a sub-industry

of IMI with higher barriers to entry and with above characteristics. To capture sub-industry specific

characteristics, we construct 7 sub-industry specific dummy variables (SID1,…,SID7) corresponding to

the 7 sub-industries (SI1,…,SI7). The sub-industry SI0 acts as the reference industry.

The method of construction of firm and year-specific TE is explained in Appendix-1 and

measurements of remaining variables are explained in the Appendix-2.

4. Period, Data and Sample

The specific time period of our study covers seven financial years (FY) 2000/01 to 2006/2007.

During this period India has become one of the most attractive destinations for FDI. There has been no

major change in policies affecting the IMI. Yet, the first 4 years of this period were characterized by

slow growth in the IMI and the remaining period was marked by a significantly higher growth

compared to the first period. Empirically, this suggests that we should control for time effect in the

proposed econometric analysis. The period of study is also important from the point of view of Indian

companies adopting better accounting standards, which has made the presentations and descriptions of

financial statements more detailed, transparent, accurate and uniform across the firms. As our study

uses firm-level data originally sourced from the annual reports of the companies, these developments

add additional feature to our study over the studies that have used data pertaining to the period prior to

the year 2000.

We obtained basic data on a number of financial and non-financial parameters for each year of

the study for designing various indicators to capture conducts and performance of a firm for carrying

out the empirical exercise. The major portion of this data and information was sourced from the

PROWESS database - an electronic database on information about the financial statements and various

other aspects of Indian firms designed by the Centre for Monitoring the Indian Economy (CMIE). Data

sourced from the PROWESS was supplemented and sometimes cross checked by obtaining relevant

24

information from additional sources and publications, namely Bombay Stock Exchange Directory,

Annual Reports of some companies, Capital Line Ole (another electronic database) or even by

personally contacting the company’s representatives in the case of some doubt on data. We also

acquired data from CMIE's Industry Market Size and Share chiefly for constructing a variable on the

index of market concentration. We also used some price deflators for which data was collected from

various publications of the Government of India (GoI). For each year of analysis, we compiled relevant

product/industry-wise data on Wholesale Price Index (base year 1993-94) from the WPI series

published by the Office of Economic Advisor (OEA), GoI. Similarly, we accessed year-wise data on

the All India Consumer Price Index Numbers (General) for Industrial Worker (base year 1982) from

the Labour Bureau, GoI. With the help of compiled data, we designed appropriate firm-level and sub-

industry level indicators.

We extracted a list of all firms belonging to the IMI available in PROWESS database. We

included all those firms in the sample for which data on each of the relevant variables were available

for at least 2 years of the 7 financial years of the study. Further, we deleted sick companies, i.e., the

companies with negative networth in a financial year, mainly with a view to remove outlier effect from

the analysis. These exclusions left us with a usable sample of unbalanced panel of 177 firms with 936

observations. The size of overall sample (as well as the size of each sub-sample of DFs and FAs) varies

from year to year during the period 2000/01 to 2006/07 of the study. As a result, we have an

unbalanced panel of firms with total number of observations over 2000/01 to 2006/07 aggregating to

936 including 261 for sub-sample of FAs and 675 for sub-sample of DFs. The number of all firms

observed in each year is lower than 177, ranging from 124 to 144 in the various years of the study.

Despite sample size being smaller than that of the PROWESS database, share of sample firms

in respect of some aspects of corporate financial indicators (say sales turnover or net worth) of the IMI

during the period of the study ranges from 66 per cent to 90 per cent depending on the individual

aspects of financial indicators. In particular, sample firms in aggregate over 2000/01 to 2006/07

covered 68 per cent of sales turnover, 90 per cent of gross profit, 85 per cent of net worth, 74 per cent

of gross fixed assets, 69 per cent of total assets, 66 per cent of foreign exchange earnings and 74 per

cent of foreign exchange outgo of all the firms belonging to the machinery industry as classified in the

PROWESS database. Considering the fact that PROWESS covers almost entire corporate sector, our

sample with such shares on the individual aspects of financial indicators can be considered as the good

representative of the corporate sector of IMI.

25

Table-1 summarises the descriptive statistics of individual variables used in the study. The

descriptive statistics include mean, standard deviations (overall, between and within), minimum and

maximum values of each variable. The table reveals that the FCD as well as all the sub-industry

specific dummy variables have no within group variation in their respective data. To know the severity

of multicolinearity problem associated with the sample, we obtained the matrix of correlation

coefficient between a pair of regressors and variance inflation factor (VIF)13. As a rule of thumb if the

pair-wise or zero-order correlation coefficient between two regressors is high, say, in excess of 0.80,

multicolinearity is considered as a serious problem (Gujarati 2004, p. 359). Again, as a rule of thumb,

if the VIF of a variable exceeds 10 that variable is deemed highly collinear (Gujarati 2004, p. 362). In

view of these rules of thumb the values of VIF and correlation coefficient did not reveal any serious

multicolinearity problem.

5. Statistical and Econometric Methods

There exists variety of methods that can be used for examining the issues set out in the

objectives of the study. The most rudimentary method involves the univariate group mean comparison

technique that compares one characteristic at a time while ignoring a large number of other

discriminant factors. Therefore, it would be appropriate to extend/enrich and compare the findings of

univariate analysis with the results obtained from multiple variable techniques. The multiple variable

techniques have advantage over univariate analysis for the former can consider an entire profile of

characteristics common to relevant firms. To classify or make predictions in situations having

dichotomous categorical dependent variable, empirical researchers have mainly employed and

estimated three types of multivariate models, namely, linear discriminant analysis (LDA), logit and

probit regression models.

5.1 Univariate Method of Analysis

The first step of this technique involves classification of an observation into one of the several a

priori groupings based on certain criteria (e.g. the two groups of FAs or DFs in our study are based on

the criterion of at least 26 per cent equity holding in a company by the foreign promoters).14 In the

second step, the value of mean and standard deviation of a variable representing particular

characteristic of a firm is calculated for the each group. Finally, a suitable statistical technique is used

13VIF shows the speed with which variances and covariance increase and can be defined as VIF = 1/(1-r223), where r2

23 is the coefficient of correlation between X2 and X3. It is called variance inflating factor because it shows how the variance of an estimator is inflated by the presence of multicolinearity. If there is no colinearity between X2 and X3 VIF will be 1. When r2

23

approaches 1, VIF approaches infinite. (Gujarati 2004, Chapter 10).

14The groups could be more than two also in a discriminant analysis.

26

for testing the significant difference in mean value of a particular variable between the two groups. The

univariate means comparison method may provide important clues about differences in conducts and

performance of FAs and DFs. To compare each aspects of conduct and performance of two groups of

firms in a univariate framework, we conduct Welch's t-test using two-samples having unequal

variances. To conduct this test we first of all need to calculate mean and standard deviation of

individual variables for each sub-sample of FAs and DFs. Thereafter, we are to obtain t-statistics with

the help of STATA software These statistics are used with t-distribution to test the null hypotheses (Ho)

for each variable that the difference in mean between the groups of FAs and DFs is zero (using a two-

tailed test) against the alternative hypothesis (Ha) that the groups have different means. In other words:

H0: mean (FA) – mean (DF) = diff = 0 against Ha: diff ! = 0

We preferred to use two-tail test because of the possibility that mean of a variable for FAs may be less

or more than that of DFs. The tests yields p-value that may (or may not) provide evidence sufficient to

reject null hypothesis.

5.2 The Empirical Models of Multivariate Analysis

The linear discriminant function used for the discriminant analysis and the empirical equations

corresponding to the theoretical models of LDA, logit and probit, as detailed in the Appendix-2, are

presented below:

Linear Discriminant Function

Z = b0 + b1GPMit + b2 TEit + b3 RDIit + b4 MTIit + b5 AMIit + b6 CAPIit + b7 SZit + b8 AGEit + b9 LEVit

+ b10 XIit + b11 MIit + b12 IMCit + b13 SID1it +,…,b19 SID7it (1)

Logit regression

Pr = E (FCDit =1| X) = 1/[1 + exp- Z] (2)

where, Z = b0 + b1GPMit + b2 TEit + b3 RDIit + b4 MTIit + b5 AMIit + b6 CAPIit + b7 SZit + b8 AGEit + b9

LEVit + b10 XIit + b11 IMIit + b12 MCit + b13 SID1it +,...,+ b19 SID7 + vit or

L = ln [Pr / (1- Pr)] = b0 + b1GPMit + b2 TEit + b3 RDIit + b4 MTIit + b5 AMIit + b6 CAPIit + b7 SZit + b8

AGEit + b9 LEVit + b10 XIit + b11 MIit + b12 IMCit + b13 SID1it +,...,+ b19 SID7 + vit (3)

Probit regression

Pr = E (FCDit =1| X) = 1- f [- (b0 + b1GPMit + b2 TEit + b3 RDIit + b4 MTIit + b5 AMIit + b6 CAPIit + b7

SZit + b8 AGEit + b9 LEVit + b10 XIit + b11 MIit + b12 IMCit + b13 SID1it +,...,+ b19 SID7 + vit)] (4)

The LDA is a statistical technique that is mainly used to classify an observation into one of the

two a priori groups dependent on the observation’s individual characteristics. Alternatively, LDA

identifies the discriminating characteristics of two groups (say FAs and DFs) of firms based on certain

27

criteria. The equation (1) is estimated for LDA. Equation 2 (or 3) and 4 represent logit and probit

models respectively in which Pr = E (FCDit =1| X) denotes conditional expectation of FCDit given X (a

vector of explanatory variables) or conditional probability that a firm will appear as FA given X. The

logit model is expressed in two forms, notably by non-linear equation (2) and linear equation (3). In

equation 3, the odd ratio Pr/(1-Pr) shows the ratio of the probability that a firm will appear as FA to the

probability that a firm will not appear as FA.

The probabilistic models (notable the logit model) are considered as the better substitutes of

discriminant analysis. Yet, the estimation results of the probabilistic models are interpreted in a slightly

different manner than that of LDA. The probabilistic models, both logit as well as probit regression

models, relate a qualitative dependent (usually dichotomous) variable to a set of continuous and/or

categorical independent variables. Probit model uses a normal cumulative distribution function (CDF),

whereas the logit model employs logistic CDF, to model such relationships between a dichotomous

dependent variable and the independent variables. In case of this study, both the models estimate the

probability of observing a firm in the group of FAs (or DFs). Thus the positive sign of the estimated

coefficient of an independent variable in these models will denote that the variable increases the

probability of the firm to appear as FA.

Each one of the three models is estimated with the help of unbalanced sample of pooled data on

individual variables used in the model. Use of pooled data set of cross-section firms over a period of

time provides us with a larger number of data points. Therefore, it increases the degrees of freedom and

reduces the co-linearity among explanatory variables and, hence, improves the efficiency of

econometric estimates. The panel data models, besides the improving efficiency of econometric

estimates, enable us to disentangle the unobserved heterogeneity (or individual effect) from the data,

which remain constant over time. Estimation of panel data models requires that there should be within

group variation in the dependent variable for adequate number of groups. Despite the superiority of

panel data models we are restricted to use only the pooled data model, as our data on FCD does not

have any within group variation. The absence of within group variation in FCD also prevented us from

using year-specific dummy variables.

6. Results

6.1 Univariate Analysis

Table-2 summarizes the results on mean, standard deviation and tests of equality of group

means of FAs and DFs with respect to 11 firm specific variables representing various firm-level

characteristics including conducts and performances. T-statistics in respect of each variable is obtained

28

by applying the formula explained in section 5.1. Thereafter, we test the null hypothesis that the

difference in mean value of each variable between the two group of FAs and DFs would be zero. The

null hypothesis is rejected in the case of 9 variables. The results indicate that FAs, as compared to DFs,

on an average achieve greater technical efficiency (TE), gross profit margin (GPM) and export

intensity. As compared to DFs, FAs spend higher portion of their revenue on research and development

as well as on import of intermediate goods and disembodied technology. As the R&D activity and use

of imported technology require higher level of skill, we may assume that skill intensity of FAs are

greater than that of DFs. These results probably suggest that FAs do have firm-specific ownership

advantage over DFs in terms of technology. In relation to DFs, FAs on an average spend less portion of

their revenue on advertising and marketing. In other words, DFs spend more towards creation of

product differentiation advantage. In comparison to DFs, FAs are also bigger in terms of their size of

their operation. Results on relative AGE and CAPI indicate that FAs and DFs do not significantly

differ in terms of years of operations and choice of technique. As compared to DFs, FAs are also found

less financially leveraged, implying that the latter finance their operations more from owned fund than

from the borrowed.

As the univariate analysis places emphasis on each individual characteristic independently from

the others, it is imperative to build upon the findings of univariate analysis and combine several

characteristics in a meaningful predictive model. The next two sub-sections of this chapter undertake

this task with the help of LDA and by estimating probabilistic models.

6.2 Linear Discriminant Analysis

Having developed various alternative multivariate models of relative characteristics of FCs and

DFs, we first examine the results obtained from LDA. LDA was performed by using SPSS – the

popular software for statistical and econometric analysis. Table-3 reports the results related to the

suitability of the LDA in three panels, A, B and C. Panel A shows that the eigenvalue and canonical

correlation statistics are 0.411 and 0.533 respectively, suggesting that the LDA model is satisfactory at

discriminating between the characteristics of FAs and DFs. The Wilks’ lambda statistics of 0.71 shows

that only 29 per cent variance in discriminant scores is explained by the differential characteristics of

FAs and DFs. The Wilks’ lambda, however, is found significant. This indicates that we can reject the

null hypothesis that the FAs and DFs have the same mean discriminant function scores and thereby

conclude that the overall model is discriminating.

Panel B assesses the contribution of each variable to the discriminant function. The results on

within group correlations between discriminant variables and standardised canonical discriminant

29

functions indicate that variables SID2, SID4, SID7, SID6, CAPI and AGE make the lowest

contributions to discriminant function. We will see later that these variables are also found statistically

insignificant in explaining the dependent variable FCD in the probabilistic models. Thus, the selecting

a variable on the basis of its level of correlations with the discriminant function may be considered as a

good criterion for its inclusion in LDA.

Panel C presents the results on the test of equality of group covariance matrices assuming

multivariate normality. Box’s M test used for this purpose shows that the null hypothesis of equality in

the group covariance matrices can be strongly rejected. This may be due either to a failure of

multivariate normality or because the group covariance matrices are not equal. In either case, the LDA

approach, together with linear classification rule is inappropriate for estimating the model. In the case

of large sample size, however, even small differences in covariance matrices may be found significant

by Box’s M when in fact no substantial problem of violation of assumptions exists. Therefore, the

researchers also look at the log of determinants of group covariance matrices. If the group log

determinants are similar, a significant Box’s M for a large sample is usually ignored. Since this study

uses the large sample, we also analyse the values of log determinants, which turn out to be dissimilar

between FAs and DFs. Thus, we cannot ignore the significance of Box’s M and therefore consider the

use of LDA less appropriate in the context of our study.

Nevertheless, we present and examine the results of LDA pertaining to the discriminating

characteristics of FAs and DFs. In particular, we examine the results of LDA following the

Mahalanobis Distance (or D) square procedure. Mahalanobis distance is the distance between a case

and centroid for each group of the dependent variable (FAs and DFs) in attribute space (defined by the

independent variable). A case will have one Mahalanobis distance for each group and it will be

classified as belonging to the group for which its Mahalanobis distance is smallest. Thus, the smaller

the Mahalanobis distance, the closure the case is to the group centroid and more likely it is to be

classed as belonging to that group. Since Mahalanobis distance is measured in terms of standard

deviation from the centroid, a case which is more than 1.96 Mahalanobis distance units from the

centroid has less than 0.05 chance of belonging to the group represented by the centroid.

In the stepwise procedure followed by SPSS, the variable that maximises the Mahalanobis

distance between the two closest groups is entered at each step. Table-4 presents the results obtained

from the LDA following the Mahalanobis D square stepwise method. As in the case of Table-3, Panel A

of Table-4 shows that model is significant but does not fulfill the criteria of equal population

covariance matrices. Focusing on the results incorporated in Panel B of Table-4, we find that out of 19

30

variables included in the model only 9 variables- TE, SZ, XI, MI, AMI, MTI, LEV, SID1 and SID5-

ultimately turn out to be significant discriminator between FAs and DFs in the stepwise procedure.

Panel C reports the values of the estimated coefficients associated with each of these variables in the

discriminant functions of FAs and DFs. We find that FAs as compared to DFs are more technically

efficient, more export intensive and more intensive in terms of import of intermediate goods and

disembodied technology. However, FAs are less leveraged and spend less as a share their sales on

advertising and marketing. FAs are also larger than DFs. It is to be noticed that LDA does not find

GPM and RDI to be a significant discriminator between FAs and DFs. On the other hand, the univariate

analysis has found GPM as well as RDI of FAs to be greater than GPM and RDI of DFs. However, both

the univariate analysis and LDA show that the AGE and CAPI are not significant discriminators

between FAs and DFs.

We will see in the next section that, despite the unsuitability of LDA for the sample used for this

study, the results obtained from LDA are more or less similar to the results found from the estimation

of probit and logit models discussed in the next sections.

6.3 Probit and Logit models

Before estimation of probit or logit models, we conducted several tests for detecting

heteroskedasticity associated with the variables used in the models. The results of these tests revealed

that the assumption of homoskedasticity is invalid. Thereafter, we estimate the probit and logit models

represented by the above equations by maximum likelihood technique with the help of STATA

software. We also obtain heteroskedasticity-corrected standard errors by following White-Huber

method with the help of robust option available in the software.

Table-5 presents the results achieved by the estimations of the logit and probit models using the

maximum likelihood methods. We may note at the outset that the estimated logit and probit models

offer similar results. The values of pseudo R2 show that both the logit and probit models achieve same

value of 0.26, implying one cannot differentiate between these models on the basis of overall goodness

of fit. The values of Wald chi2 and corresponding p-value of zero suggests that the each (probit as well

as logit) model as whole is statistically significant, as compared to model with no regressors. Thus,

there is little to choose between probit and logit approaches.

The results on firm-specific variables show that the coefficients of GPM, AGE, CAPI and RDI

are statistically insignificant. On the other hand, the coefficients of TE, SZ, XI, MI and MTI are

positive and significant and coefficient of LEV and AMI are negative and significant in both the

models. Comparing the results of univariate analysis and LDA against the results of probabilistic

31

models, we find that: a) GPM and RDI differ significantly between FAs and DFs in univariate analysis,

while both are not found as discriminating factors between FAs and DFs in LDA. GPM and RDI also

do not impact the probability of a firm to appear as FA in the presence of other variables in the both

types of the probabilistic models; b) AGE and CAPI do not differ significantly between FAs and DFs

in the univariate analysis, LDA and the probabilistic models; c) the signs of the statistical significant

coefficients of TE, SZ, XI, MI, MTI, AMI, LEV are identical in every types of analysis; d) two sub-

industry specific dummy variables, SID1 and SID5 appear as significant discriminators between FAs

and DFs in LDA, but the coefficients of the same are observed statistically insignificant in both the

probabilistic models.

As discussed earlier, the multivariate analyses based on probabilistic models are considered

more appropriate and theoretically sound, we thus consider the results obtained from the probabilistic

model to be the final. We therefore discuss these results elaborately and draw conclusions and policy

implications from the same. The estimation results of probit model on the factors that influence the

probability of being a firm in foreign ownership also gives marginal effects (Table-5). The marginal

effects are calculated for discrete change of dummy variable from 0 to 1 at the sample means and

measured in terms of absolute value of a coefficient. Among the statistically significant explanatory

variables, the MTI has the greatest effects followed by AMI, MI, TE, XI, LEV and SZ in descending

order.

Intensity of imported disembodied technology (MTI) with the highest positive marginal effect

indicate that the likelihood of being FA is the greatest for a firm that makes higher payment (as a ratio

of its sales) for import of foreign disembodied technology. These results are in line with the findings of

several Indian studies cited in the section-2 [e.g. Siddharthan and Krishna 1994, Basant 1997, Kumari

2007, Ray and Rahman 2006, Ray and Venaik 2001 & 2008]. It is paradoxical that the firms with FDI,

which are supposed to receive advance foreign technology from the MNE system at no or nominal

cost, are spending substantially higher amount (than DFs) on arm's length purchase of foreign

technology. In addition, the higher intensity of payment for import of disembodied technology by FAs,

coupled with no difference in R&D intensity of FAs and DFs, imply that FAs not only rely more on

foreign technological know-how but also do not make major attempts to adapt or absorb the imported

technology. As FAs normally buy technology from their own MNE system, it seems that they over pay

for the same in the intra-firm transactions for appeasing their parents. However, this issue needs further

investigation which is beyond the scope of this thesis.

32

The second most important factor explaining probability of a firm to be in foreign ownership is

the advertising and marketing intensity (AMI). The significantly negative coefficients of AMI observed

in the estimated probit and logit models show that the FAs spend less for creating product

differentiation advantage than DFs. This result probably indicates that the already established

international image of MNEs and brand equity of their products are requiring FAs in the IMI to spend

less on current advertising and marketing for creating product differentiation advantages. Besides, the

FAs gain from the spillovers of the worldwide advertisements of their respective MNEs but do not

contribute monetarily for the same. Another reason cited by Ray and Rahman (2006), who do not find

product differentiation to be a discriminator between FAs and DFs, is that the threat from the entry of

large number of MNEs after liberalisation from 1991 has forced oligopolistic DFs to spend heavily on

advertising and marketing for the protection of their market share.

The third factor is intensity of imported intermediate goods (MI). The combined results on

higher intensities of import for intermediate good and disembodied technology can also be interpreted

as evidence of MNEs' indulgence in intra-firm trade at transfer prices for boosting their global profit.

Inflation of payment on royalty and technical fee by FAs has been used as good means for reducing

local taxes in the host country and transferring earned profit out of the host country (Lecraw 1983,

Bellak 2004a). Our finding on MI is in line with the recent findings in the Indian studies [e.g. Ray and

Rahman 2006, Ray and Venaik 2001 & 2008].

Our finding on technical efficiency (TE) is consistent with the prediction of internalisation (or

transaction cost) approach of FDI and findings of several empirical studies that the FAs are more

productive/efficient than the DFs as cited in the Section-2 [viz. Sinha (1993), Kathuria (2001, 2002),

Ray (2004), Goldar et al. (2004), Sasidharan and Ramnathan (2007) for Indian manufacturing sector].

Insignificance and significant coefficients corresponding to GPM and TE in multiple variable analyses

suggest that the latter captures the essence of performance in a better way and thereby act as the

discriminator between FAs and DFs. Besides, the combined results on GPM and TE can be interpreted

in the following manner. First, MNEs do not acquire firms with monopoly profits reflected as reflected

by GPM but acquire technically efficient firms having potential for giving consistently average profits

as applicable to the industry. Secondly, the earlier studies that have found profitability to be the

important discriminant between DFs and FAs have not used any measure of productivity or efficiency

as an additional variable in their multi-variate analysis.15 Therefore, their measures of profitability

might have also captured the effect of technical efficiency. Thirdly, the motivation of MNEs to

15See for example Kumar (1991).

33

minimise their tax burden and thereby show reduced profit, than they actually earn, may be responsible

for understating the profits of FAs and thereby the statistical insignificance of the coefficients of GPM

in LDA and probabilistic models. This is termed by Bellak (2004) as the accounting factor.

Presumably, FAs are better at minimizing their tax burden than DFs owing to the formers’ affiliations

to the MNE system having better expertise and opportunities in terms of multinational network.

Fourth, if opportunity cost of internally generated funds (i.e. retained earnings) are lower than that of

externally raised funds, the managers of FAs will accept lower profitability when they use reinvested

profits (Bellak 2004).

Significant and positive coefficient of export intensity (XI) suggests that FAs are not only

concentrating on the host country market, but also by using the India's locational and comparative

advantages, have gained relative competitive advantage over DFs on the export front. This finding is

consistent with the findings of the larger set of recent Indian studies [e.g. Siddharthan and Nollen

(2004), Chhibber and Majumdar (2005) and Rasiah and Kumar (2008)]. However, our study

contradicts the findings of Ray and Rahman (2006), employing LDA, in this respect.

FAs are also found less financially leveraged than DFs, indicating that the FAs use greater

amount of internal funds for financing their operations. This support our hypothesis build on the

arguments favouring lower financial leverages maintained by MNEs as compared to DFs and empirical

evidences on the same from various studies mentioned in section-2. Thus, our study is in line with the

finding of the majority of empirical studies which report FAs to be maintaining lower financial

leverage than the DFs in the context of the developed countries [see e.g. Akhtar and Barry (2009) for

Japan; Lee and Kwok (1988), Burgman (1996), Homaifer et al. (1998), Chen et al. (1997), Chkir and

Cosset (2001) and Doukas and Pantzalis (2003) for USA].

Size of the firm, generally reflecting the firm’s ownership of financial and non-financial

resources, has positive influence on the firms’ probability to appear in the group of FAs (than DFs).

Besides, the LDA also shows that the FAs have larger size than DFs. As discussed in section-2, the

reason for this could be found in the FAs’ ownership of higher amount of financial and non-financial

resources to overcome sunk and transaction costs associated with doing business in a foreign location.

Our finding on firm size is similar to the findings of studies for East Asian countries [Ramstetter

(1999a); Takii and Ramstetter (2003)].

The coefficients of IMC turn out to be insignificant in the estimated probit as well as logit

model. This indicates that the probability of a firm’s appearance in the group of FAs (or DFs) is not

dependent on the market concentration. Similarly, the coefficients of none of the sub-industry specific

34

dummy variables are found statistically significant either in the estimated probit or logit model. These

results hint that the FAs do not show any preference for locating in one or other sub-industries of IMI.

This might have happened because the sub-industries of the IMI may not be differing sufficiently in

terms of overall index of characteristics so as to warrant the special attention of MNEs.

7. Summary and Conclusions

The objective of this paper was to empirically examine the differences in the relative

characteristics, conducts and performance of two ownership groups of firms, FAs and DFs, in the IMI

during 2000/01 to 2006/07. For this purpose, we first applied the univariate statistical method based on

Welch's t-test for comparing the mean value of a variable between two groups of firms. The findings of

the univariate analysis revealed that the FAs (as compared to DFs) have significantly greater TE, GPM,

SZ, RDI, XI, MI, and MTI. However, in relation to DFs, FAs as a group is found on an average less

advertisement and marketing intensive (AMI) as well as less financially leveraged (LEV). T-test did

not reveal significant differences between FAs and DFs in terms of the mean values of AGE and CAPI.

Since the univariate method compares one characteristic at a time, we conducted the

multivariate LDA. We found that the results of LDA were similar to the univariate analysis in respect

of all the firm-specific variables except in the case of GPM and RDI, as both the variables did not

discriminate between FAs and DFs in the presence of other variables. However, the LDA is based on

certain restrictive assumptions (viz. normality and equal group covariance matrix in respect of the

independent variables), which were found inconsistent with the sample used for this study.

We, therefore, estimated two dichotomous logit and probit models, which do not require the

fulfillment of these assumptions. The estimation results of both types of the probabilistic model were

identical and almost similar to that of LDA. However, the interpretations of the results obtained from

the probabilistic models somewhat differ from that achieved by employing LDA. The findings of the

probabilistic models indicated that the likelihood of the firms to fall in the category of FAs is higher if

they have the greater technical efficiency (TE), firm size (SZ), export intensity (XI), intensity of import

of intermediate goods (MI) and intensity of import of disembodied technology (MTI) but the lower

advertisement and marketing intensity and financial leverage (LEV). The result on AMI probably

indicates that the previously established international image of MNEs and brand equity of their product

are causing FAs to spend less towards advertising and marketing for creating product differentiation

advantages vis á vis DFs. The result on financial leverage implies that FAs adopt prudential practice of

employing greater amount of internal funds (in relation to debts) for financing their operations. Results

35

on relative AGE and CAPI indicate that FAs and DFs do not significantly differ in terms of years of

operation and capital intensity of production in the IMI.

In view of the common significant findings of the multivariate analyses from the LDA and

probabilistic models in respect of the several variables, we conclude that our empirical analysis

supports the proposition that the FAs and DFs differ in terms of the many aspects of conducts and

performance in the IMI. These findings also give some indications about the quality of FDI that has

come to the IMI during the aftermath of economic reforms. First, it seems that the superior resources

and capabilities of FAs confer them higher technical efficiency (but not overall performance or the

monopoly power in terms of GPM) and export intensity in relation to DFs. Second, as the intensity of

import of intermediate goods in FAs is significantly higher than that of DFs, the former group tends to

have fewer linkages with domestic suppliers of intermediate goods including capital goods, raw

material, components and spare parts. In other words, DFs with their activities in the IMI are providing

higher linkages with the indigenous suppliers. Third, the combined results on higher expenses on

import of intermediate goods and import of foreign technology by FAs and no difference in gross profit

margins between FAs and DFs point out that the FAs are probably engaged in the transfer of profits to

the MNE system through intra-firm trade. This aspect, however, require further research which is

beyond the scope of this study. Fourth, despite the higher import of intermediate goods and

disembodied technologies, FAs are not spending higher amounts on R&D towards adaptation or/and

absorption of the imported technology and indigenization of the imported inputs. As a result, R&D

intensities of FAs and DFs are the same. Finally, FAs have no preference for locating themselves in one

or the other sub-industries of the IMI, indicating that the sub-industries of machinery industry do not

differ significantly among one another in term of locational attractiveness in the eyes of MNEs.

36

Table-1: Descriptive Statistics of Variables for full Sample, 2000/01-2006/07

Variable Mean Std. Dev. Min Max

FCD overall 0.2788 0.4487 0.0000 1.0000

between 0.4301 0.0000 1.0000

within 0.0000 0.2788 0.2788

TE overall 0.7096 0.0816 0.5377 0.9934

between 0.0838 0.5447 0.9932

within 0.0028 0.7025 0.7156

GPM overall 0.1904 0.1173 -0.4871 0.7081

between 0.0979 -0.1754 0.4736

within 0.0683 -0.2759 0.6389

SZ overall 3.4278 1.6245 -0.1372 8.8828

between 1.5575 0.2772 8.5254

within 0.2773 2.1015 4.9944

AGE overall 3.1944 0.7298 0.0000 4.6250

between 0.7373 0.8959 4.6000

within 0.1266 2.0978 3.8896

CAPI overall 4.7216 5.0334 0.2844 50.0000

between 5.0590 0.3259 39.5469

within 1.2665 -4.5606 15.1747

AMI overall 0.0309 0.0333 0.0000 0.2506

between 0.0314 0.0000 0.2197

within 0.0127 -0.0548 0.1597

MTI overall 0.0031 0.0074 0.0000 0.0743

between 0.0060 0.0000 0.0372

within 0.0040 -0.0215 0.0547

RDI overall 0.0035 0.0060 0.0000 0.0398

between 0.0053 0.0000 0.0284

within 0.0027 -0.0093 0.0260

LEV overall 0.3338 0.2526 0.0000 0.9863

between 0.2432 0.0000 0.9577

within 0.1070 -0.1947 0.7288

XI overall 0.1247 0.1736 0.0000 0.9922

between 0.1523 0.0000 0.7551

within 0.0886 -0.3857 0.6732

MI overall 0.0930 0.1027 0.0000 0.5823

between 0.0918 0.0000 0.4633

within 0.0455 -0.1904 0.4421

IMC overall 0.4038 0.1596 0.1256 0.8955

between 0.1523 0.1580 0.7762

within 0.0568 -0.0171 0.6845

SID1 overall 0.1378 0.3449 0.0000 1.0000

between 0.3550 0.0000 1.0000

within 0.0000 0.1378 0.1378

SID2 overall 0.0951 0.2935 0.0000 1.0000

between 0.2955 0.0000 1.0000

within 0.0000 0.0951 0.0951

SID3 overall 0.0652 0.2470 0.0000 1.0000

between 0.2521 0.0000 1.0000

within 0.0000 0.0652 0.0652

SID4 overall 0.1229 0.3285 0.0000 1.0000

between 0.3243 0.0000 1.0000

within 0.0000 0.1229 0.1229

SID5 overall 0.1816 0.3857 0.0000 1.0000

37

between 0.3812 0.0000 1.0000

within 0.0000 0.1816 0.1816

SID6 overall 0.0823 0.2749 0.0000 1.0000

between 0.2955 0.0000 1.0000

within 0.0000 0.0823 0.0823

SID7 overall 0.2404 0.4275 0.0000 1.0000

between 0.4231 0.0000 1.0000

within 0.0000 0.2404 0.2404

38

Table-2: Comparing Characteristics of FAs and DFs-Univariate Method

(Tests of Equality of Group Means)

Variable

Domestic Firms Foreign Controlled Firms

Tests of

Equality of

Group

Means

Obs. Mean Std. Dev. Obs. Mean Std. Dev.Welch's

d. o. f.

TE 675 0.6976 0.0777 261 0.7405 0.0835 445.23

GPM 675 0.1800 0.1187 261 0.2175 0.1094 511.39

SZ 675 3.1821 1.6779 261 4.0633 1.2766 619.45

AGE 675 3.1911 0.7251 261 3.2028 0.7431 463.90

CAPI 675 4.7699 5.5087 261 4.5967 3.5243 713.20

AMI 675 0.0331 0.0347 261 0.0254 0.0287 568.06

MTI 675 0.0016 0.0052 261 0.0070 0.0104 312.36

RDI 675 0.0032 0.0058 261 0.0043 0.0065 427.06

LEV 675 0.3655 0.2498 261 0.2516 0.2415 489.15

XI 675 0.1131 0.1744 261 0.1548 0.1683 489.91

MI 675 0.0705 0.0873 261 0.1513 0.1159 380.61

Note: * and ** denote significance levels at 1% and 5% respectively

39

Table-3: Results of LDA

Panel A: Canonical Distance Function

Eigenvaluea Canonical

Correlationb Wilks’ Lambdac χ2 (19) Prob > χ2

0.411 0.533 0.709 318.173 0.000

Panel B: Pooled Within-Groups Correlations between Discriminant Variables and Standardised Canonical

Discriminant Functions

(Variables ordered by the absolute size of correlation within function)

Var Corr Variable Corr Var Corr

MI 0.588 SID3 -0.167 SID4 -0.069

MTI 0.540 AMI -0.162 SID7 -0.041

SZ 0.391 IMC -0.139 SID6 0.034

TE 0.378 SID1 0.130 CAPI -0.024

LEV -0.322 RDI 0.128 AGE 0.011

GPM 0.277 SID5 -0.120 - -

XI 0.169 SID2 0.078 - -

Panel C: Test of Equality of Group Covariance Matrices Using Box’s M

FCD Rankd Log determinantd

0 19 -67.985

1 19 -71.062

Pooled within-groups 19 -67.244

Test Results (tests null hypothesis of equal covariance matrices)

Box’s M Approximate F (190, 813887.4) Prob > F

1491.554 7.632 0.00

Notes:

a. Eigenvalue = between group sum of squares/within group sum of squares

b. Canonical correlation = (between group sum of squares/total sum of squares)1/2. In the two group case, the canonical

correlation is the correlation coefficient between the discriminant score and the group variable.

c. Wilks’ lambda = within group sum of squares/total sum of squares. Wilks’ lambda captures the proportion of the total

variance in the discriminant scores not explained by the differences between the groups.

d. The ranks and natural logarithms of determinants are those of group covariance matrices.

40

Table-4: Results of LDA based on Stepwise Procedure

d.

Panel A: Canonical Distance Function

EigenvalueCanonical

CorrelationWilks’ Lambda χ2 (19) Prob > χ2

0.394 0.532 0.717 308.666 0.000

Panel B: Test of Equality of Group Covariance Matrices Using Box’s M

FCD Rank Log determinant

0 9 -37.027

1 9 -36.572

Pooled within-groups 9 -36.395

Test Results (tests null hypothesis of equal covariance matrices)

Box’s M Approximate F (45, 864988.6) Prob > F

471.810 10.344 0.00

Panel C: Mahalanobis D Squired Stepwise LDA

Variable

Entered

Mahalanobis D Squired

Statistics between FAs and DFs

Exact F

Statistic df1 df2

1 MI 0.706 132.867 1 934.000

2 MTI 1.241 116.635 2 933.000

3 LEV 1.428 89.427 3 932.000

4 SZ 1.567 73.485 4 931.000

5 TE 1.659 62.193 5 930.000

6 AMI 1.804 56.279 6 929.000

7 SID1 1.873 50.037 7 928.000

8 XI 1.919 44.812 8 927.000

9 SID5 1.954 40.524 9 926.000

Notes: a) At each step, the variable that maximizes the Mahalanobis distance between the two closest groups is entered; b)

Maximum number of steps is 38; c) Minimum partial F to enter is 3.84; d) Maximum partial F to remove is 2.71; e) F level,

tolerance, or VIN is insufficient for further computation.

Panel-D: Discriminant Functions of FAs and DFs

Cat TE SZ XI MI AMI MTI LEV SID1 SID5 Const

DFs 127.526 1.013 5.997 -10.357 -42.088 32.387 15.655 0.691 -1.107 -48.517

FAs 133.223 1.196 7.418 -2.438 -53.502 135.385 14.509 1.267 -1.643 -55.05

41

Table-5: Logit and Probit Models: Estimation Results

Logit Model Probit Model-1 Probit Model-2

Explanat

ory

Variable

Coef.

Het.

corr.

Std. Err.

z-stat Coef.

Het.

corr.

Std. Err.

z-stat dF/dx

Het.

corr.

Std. Err.

TE 6.185 1.324 4.67* 3.728 0.761 4.90* 1.120 0.228

GPM -0.251 1.042 -0.24 -0.205 0.579 -0.35 -0.062 0.174

SZ 0.194 0.076 2.55* 0.122 0.042 2.90* 0.037 0.013

AGE -0.004 0.004 -0.99 -0.003 0.002 -1.35 -0.001 0.001

CAPI -0.030 0.019 -1.64 -0.016 0.011 -1.51 -0.005 0.003

AMI -10.645 2.811 -3.79* -6.383 1.586 -4.02* -1.918 0.481

MTI 81.725 16.623 4.92* 44.564 9.347 4.77* 13.394 2.871

RDI -11.963 13.775 -0.87 -7.511 8.229 -0.91 -2.257 2.476

LEV -1.321 0.450 -2.93* -0.669 0.240 -2.79* -0.201 0.071

XI 1.140 0.555 2.05** 0.757 0.301 2.51* 0.228 0.091

MI 6.525 1.055 6.19* 3.675 0.594 6.19* 1.104 0.181

IMC -0.727 0.875 -0.83 -0.518 0.436 -1.19 -0.156 0.132

SID1 0.295 0.560 0.53 0.103 0.279 0.37 0.032 0.088

SID2 0.286 0.570 0.50 0.127 0.287 0.44 0.040 0.093

SID3 -0.551 0.648 -0.85 -0.299 0.316 -0.95 -0.081 0.075

SID4 0.042 0.468 0.09 -0.062 0.246 -0.25 -0.018 0.071

SID5 -0.656 0.522 -1.26 -0.438 0.264 -1.66 -0.117 0.063

SID6 -0.181 0.555 -0.33 -0.060 0.287 -0.21 -0.018 0.083

SID7 -0.001 0.494 0.00 -0.034 0.247 -0.14 -0.010 0.073

Const. -5.811 1.146 -5.07* -3.423 0.636 -5.38

Number of observations 936 Number of observations 936

Wald Chi2 (19) 193.88 LR Chi2 (19) 228.39

Prob > chi2 0.00 Prob > chi2 0.00

Pseudo R2 0.26 Pseudo R2 0.26

Log likelihood -407.56 Log likelihood -408.63

Note: *, ** denote level of significance at 1 per cent and 5 per cent per cent respectively.

: dF/dX is for discreet change of dummy variable from 0 to 1

42

Appendix-1

Technical Efficiency

To calculate firm and year specific TE, we estimate SFPF model by adopting Battese and

Coelli's (1992) specification involving the use of unbalanced panel data. Our empirical model consists

of a single equation production function with natural logarithm of output as the dependent variable, and

material input, labour input, capital input as three independent variables. The Cobb-Douglas form of

production function is chosen, because of its well-known advantages and simplicity. In principal,

confining the analysis to this one functional form can be somewhat restrictive. However, a few studies

[e.g. Kopp and Smith (1980) and Krishna and Sahota (1991)] suggest that the functional specifications

have small impact on measured efficiency. In a relatively recent study, Driffield and Kambhampati

(2003) do not find significant differences in the estimation results obtained either from trans-log or

Cobb-Douglas specification. The log linear form of Cobb-Douglas production function to be estimated

in accordance with the estimation methods described above is expressed as follows:

ln Yjit

= b0 + b1 ln Mjit + b2 ln Ljit + b3 ln Kjit+ Vjit – Ujit (6)

where Y, M, L, K represent output, material input, labour input and capital input respectively. The

subscript j (j = 1,…,177) refers to the jth sample firm; i (i = 1,…,936) denotes ith observation and t (t =

1,…,7) represent year of operation. The ln symbolises natural logarithm. Vjt's are assumed to be

independently and identically distributed (iid) as N(0, v2) reflecting two-side “statistical noise”

component that accounts for the effect of all random factors such as the measurement error, luck,

machine performance, etc.; Vjt are also assumed to be independent of Ujt and the input vector Xjt; Uj's

are non-negative random components assumed to be iid as non-negative truncations of the N(,u2)

distribution; Uj's are assumed to capture technical inefficiency in production, since the non-negative

assumption of U ensures that the firm’s actual production point lies beneath the stochastic frontier and

the gap between the point frontier and actual point thus measures technical inefficiency; Eta () is an

unknown scalar parameter to be estimated, reflecting the time trend of the efficiency of individual

firms. We use Coelli's (1996) “FRONTIER 4.1” software for estimating above equation by MLE

method and thereafter obtaining the parameters of the model and predictors for the year-specific and

firm-specific TE. In this framework, TE of a given firm (in a given year) is defined as the ratio of its

mean output (conditional on its level of factor inputs and firm effects) to the corresponding mean

output if the firm utilizes its levels of inputs most efficiently (Battese and Coelli1992).

43

We use a set of unbalanced panel data on a sample of 177 firms belonging to IMI. We consider

the data for 7 years during the period 2000/01 to 2006/07. A total of 936 observations are used as 303

observations are missing from the panel. The data on nominal value of each of the variables employed

to represent output and inputs of a firm is collected from the Prowess database for each year of the

study. These data include: a) value of production (VoP) that is rupee value of net sales plus net increase

or decrease in stock of finished goods, b) aggregate annual expenses incurred by a firm on the purchase

of raw materials, components, stores, spare parts, etc. It also includes expenses incidental to the

purchase of materials, c) wage bill i.e. a firm’s annual staff expenses on payment of wages and salaries,

bonus, contribution to and provision for provident, pension, gratuity funds, etc. and d) the original cost

of plant and machinery as at the end of a financial year.

Since we use many years of data on a firm, we need to compute real values of the same by

deflating the value of each input and VoP by the appropriate annual price indices. Hence, we obtained

relevant product-wise data on Wholesale Price Index (WPI) for each year of the study from the WPI

series published by Office of Economic Advisor (OEA), Government of India. To deflate data on wage

bill, we collected data All India Consumer Price Index Numbers (General) for Industrial Worker (CPI)

from Labour Bureau, Government of India. In the following paragraphs, we discuss the method of

constructing each variable employed for estimation of stochastic frontier production function. In

addition, we also explain the justification for and limitation of data utilized for measuring output and

input variables.

Output (Y): WPI deflated VoP represents the output (Y) of a firm in our study. To deflate VoP, we have

used year-wise data on WPI for a firm's major product group. In this regard, the major product group of

each company was matched with the WPI classification, and the matching price series was chosen for

the deflation. If the appropriate deflator was not available, the deflator corresponding to the nearest

product group is utilized for the purpose. For a few diversified companies operating in various

segments of machinery industry, we have used WPI of machinery industry as the deflator. The value of

production, instead of value added, is employed to measure the output because: (i) the use of the former

facilitates the inclusion of material input as another important input of production, that can also be used

efficiently (or inefficiently) along with the labour and capital, (ii) the use of value added as a measure

of output can yield misleading results if there is imperfect competition or increasing returns to scale

(Basu and Fernald 1995). Moreover, the option to employ value added or value of production depends

upon whether there are substantial gains in the management and procurement of raw material to firms,

and thereby it is essentially an empirical question (Patibandala 1998 and Driffield and Kambhampati

44

2003). Many Indian studies in recent years have estimated production function with material input as

an important independent variable (see e.g. Driffield and Kambhampati 2003 and Banga 2004).

Material Inputs (M): Materials inputs (M) constitute one of the important constituents of production in

the business. To remove the effect of year-to-year change in prices, M in this study is deflated by WPI

corresponding to the main product group to which M belonged. For this purpose, M of each company

was divided into various categories and matched with the WPI classification, and the best available

price series was chosen for deflation.

Labour Input (L): Labour input is measured by "man hours", "workers", "number of employees".

Indian firms rarely report this information in their annual reports, since the Indian Company Law does

not make it mandatory. Besides, it is normal practice among Indian firms, particularly in the recent years,

to outsource a number of manual works to labour contractors. The payments made to labour contractors

are included in the wage bill of the firm but the workers employed through the contractors are not

included in the payroll of the firm. In view of these, we employ total wage bill, which also reflects the

skill composition of employees at firm level (Bhavani and Tendulkar 2001), as a proxy measure for the

labour input in our study. Some scholars in India have preferred to use wage bill as the measure for

labour input in their respective studies (see for example Siddharthan and Lal 2004, Ray 2006). As we

use panel data, we deflate WS by the CPI, so as to mitigate the effect of inflation on the wage bill of a

firm.

Capital Input (K): Ideally, capital input (K) should be measured by the current replacement cost of the

fixed assets of a firm. Nevertheless, the absence of relevant information/data has compelled the

researchers to follow alternative methods for measuring capital input in their empirical studies. One

such widely used method captures K by the gross (or net) fixed assets of a firm either in nominal term

as given in the annual report of a firm or more satisfactorily in real term, which is calculated as gross

(or net) fixed assets deflated by an appropriate price index. We also follow the similar method. To

capture K, however, we utilise data only on the original cost of plant and machinery (or gross fixed

stock of capital), rather than the gross fixed assets that includes the original cost of land and building

as well. We exclude land and building from the gross fixed assets as many companies use rented

premises and the value of land can be significantly under (or over) estimated in the Indian conditions.

We do not use data on net fixed cost of plant and machinery because many Indian companies

manipulate data on depreciation and machineries are used even beyond their life span.

45

Results of maximum likelihood estimates of parameters of SFPF are given in Table-A.1. The

results show that the coefficients of each of the three inputs explaining production behaviour of sample

firms are statistically significant. In our model, ML estimates of coefficients also signify elasticity of

output with respect to material, labour and capital input. The comparison of these elasticity show that

elasticity of output with respect to material input (0.71) is the highest and substantial, followed by

elasticity of output with respect to labour (0.14) and capital input (0.10) respectively. Although the

value of the coefficient associated with material input is substantial, it is much less than the unity.

Notably, when we use two input production function, ignoring raw material, we implicitly assume that

the coefficient associated with material input is close to unity. Further, return to scale, measured as a

sum total of these elasticities (0.95), is quite close to unity, indicating that the production technology is

characterised by constant returns to scale.

The software also gave the firm specific and year-specific TEjt from which we calculated the

mean value of TEjt over the data period of a firm. The analysis of this data shows the most technically

efficient firm with mean TE of 99.3 per cent belongs to the group of FAs whereas the least technically

efficient firm with mean TE of 55.5 per cent belongs to the group of DFs; b) the five most technically

efficient firms in the sample includes two FAs, each one with mean TE of 99.3 per cent and 97.0 per

cent, and three DFs, each one with mean TE of 96.3 per cent, 96.1 per cent and 95.9 per cent; c) the

five least efficient firms, with mean percentage TE of 57.9, 57.9, 55.8, 55.1 and 54.5, belong to the

group of Dfs.

Table-A.1: Maximum Likelihood Estimates of Parameters of SFPF

Variable/Parameters Coefficient t-ratio

Ln M 0.7059 85.68*

Ln W 0.1399 8.13*

Ln C 0.1004 6.83*

Constant 1.2017 29.17*

Sigma-squared 0.0315 5.62*

Gama 0.7765 32.13*

Mu 0.3127 9.44*

Eta 0.0064 0.8357

Log likelihood function 705.57

Number of cross-section 177

Number of Years 7

Number of Observations 936

Number of Observations not in the panel 303

Note: * shows that the coefficient is significant at one per cent level.

46

Appendix-2

Construction and Measurement of Other Variables

FAs, DFs and FCD: We adopted an appropriate and objective criterion for segregating sample firms

into two ownership groups, FAs and DFs. This criterion was mainly based on certain provisions of the

Indian Company Act 1957, which states that an investor can block special resolution in a company by

holding a minimum of 26 per cent of equity in the paid-up share capital of a public limited company.

Following this criterion, we defined a sample company as FA if a foreign promoter holds at least 26 per

cent share in the paid-up capital of the company. Accordingly, DF is referred as a company having less

than 26 per cent equity by a foreign promoter. A further checking on the FAs revealed that each one of

them had affiliation with a reputed MNE. FCD assumes value 1 for a FA and 0 for a DF.

Capital Structure (LEV): In the empirical research, two ratios are normally utilised to measure

leverage: (i) long-term debt to total debt plus market value of equity and (ii) long-term debt to long-

term debt plus market value of equity. In this study, we specifically measure the leverage of a firm by

the ratio between the medium and long-term debts and net worth. The medium and long term debts of a

company include the debt of over one year maturity. Net worth is the summation of equity capital and

reserves and surplus. In the reserve and surplus, we do not include revaluation reserves. We represent

this ratio by LEV, higher LEV of a firm (relative to other firms) means that it is financing greater

proportion of its assets by debt than by owned fund (i.e. net worth).

Firm Size (SZ): Sales turnover is a most commonly used measure of firm size in empirical studies on

manufacturing sector. We approximate sales turnover by net sales (NS), which equals gross sales minus

indirect taxes. NS does not include other income from non-recurring transactions, income of extra-

ordinary nature and prior period income. We follow this concept but measure firm size (SZ) by natural

logarithmic value of net sales of a firm in a year. This measure of firm size has advantage over

measuring size by absolute value of net sales as the former reduces degree of variability in size across

firms, and thereby avoids the problem of heteroskedasticity in the estimation of the regression

equations.

Age of a Firm (AGE): Age of a firm is measured by the number of years of operation of a firm which

is the difference between the year of presence in the sample and the firm’s year of incorporation to. As

every year of operation may not add significantly to the experience or oldness, we use natural

logarithm of age (AGE) to represent the age of a firm.

Capital intensity (CAPI): Capital intensity (CAPI) is measured by the ratio of the cost of plant and

machinery to wage bill of a firm in a year.

Product Differentiation (AMI): We measure product differentiation advantage of a firm by its

advertising and marketing intensity (AMI), which the ratio of sum of a firm's expenditure on

advertising and marketing to net sales in financial year. The advertising expenses include expenses on

launching, promotion and publicity of goods, etc. and marketing expenses comprises commission paid

to selling agents, discounts, rebates, etc.

Export Intensity (XI): It is a ratio of export to net sales of a firm in a FY in which export is measured

by the firm’s earnings from the f.o.b. value of exports of goods and services.

Intensity of Imported Intermediate Goods (MI): MI is a ratio between c.i.f values of imported inputs

to net sales of a firm in a FY. The imported inputs include raw material, stores, spare parts, capital

goods, etc. We use combined value of imported inputs as some firms do not report reliable data on

import of capital goods and raw materials separately and also both the components of imports provide

benefits of foreign networks for exports.

47

Intensity of Imported Disembodied Technology (MTI): Indian firms import disembodied technology

from a foreign technological collaborator against the payment of royalty and technical fee and /or

lump-sum payments for obtaining technical know-how, use of patents, engineering services, drawings

and designs, brand names, trademarks and the like, etc. The royalty is normally paid on the recurring

basis as a certain percentage of domestic sales and/or of exports while technical fee may be paid on

lump-sum basis as one-time payments. The sum of royalty (net of tax) and lump sum payments may

approximate that part of technological capability of a firm, which is acquired by the import of

disembodied technology. We measure intensity of imported disembodied technology of a firm by the

ratio of sum of royalty and lump sum payment to net sales.

Profitability (GPM): We capture profitability by a measure of gross profit margin (GPM) that is a

ratio of gross profit-to-net sales. The numerator gross profit is defined as profit before depreciation,

interest, lease rental and direct taxes.

Financial Capacity (FINC): FINC is measured by a ratio of networth to total assets of a firm.

Sub-industry Specific Dummy Variables: We categorise our sample firms into its 8 major sub-

industries of the IMI, namely, SI0, SI1,…,SI7. Thereafter, we construct 7 dummy variables, SID1,

…,SID7, corresponding to 7 sub-industries SI1,…,SI7. The observations on a dummy variable (say

SID1) assumes the value 1 if a sample firm belongs to the corresponding sub-industry (say SI1),

otherwise 0. The sub-industry SI0 is treated as the reference sub-industry, therefore, we do not use

dummy variable for this sub-industry so as to avoid dummy variable trap.

Index of Market Concentration (IMC): In order to construct IMC, we first categorise the IMI into 8

sub-industries (SI1,….,SI8) with the help of facilities provided in PROWESS. A minimum 51 per cent

of gross sales made up from a sub-industry in a particular financial year is used as the norm for this

reclassification. IMC is calculated as the sales weighted average of an index of a four-firm seller

concentration ratio (SCR4) of each of the sub-industries of IMI in which a firm operates. The SCR4 is

defined as the share of sales of four largest firms taken together in gross sales of a sub-industry of

machinery industry. Since a sample firm may operate in one or multiple sub-industries belonging to

machinery industry, we calculate a weighted average of SCR4 to obtain firm-specific IMC. The weight

is calculated as ratio of a firm's sales revenue generated from an individual sub-industry to gross sales

of the firm in a year. The procedure of calculating IMC can be more clearly illustrated by the following

example. If a firm's gross sales of Rs.15 crore generated from sale of Rs.10 crore worth of bearings

(SCR4 = 0.90) and Rs. 5 crore worth of pumps (SCR4 = 0.30), IMC applicable to the firm would be

0.70 (10/15*0.90 + 5/15*0.30).

Year-specific Dummy Variables: To account for developments over the period of study, we employ

six year-specific additive dummy variables, YD02, YD03, YD04, YD05, YD06 and YD07

corresponding to the years 2001/02, 2002/03, 2003/04, 2004/05, 2005/06, 2006/07. The dummy

variable YD02 takes value 1 for the year 2001/02 and 0 for other five years; YD03 assumes value 1 for

the year 2002/03 and 0 for other five years; YD04 takes value 1 for the year 2003/04 and 0 for other

five years; YD05 takes value 1 for the year 2004/05 and 0 for the other five years; YD06 takes value 1

for the year 2005/06 and 0 for other five years; YD07 takes value 1 for the year 2006/07 and 0 for other

five years. We do not use any dummy variable for the reference year 2000/01 to avoid dummy variable

trap.

48

References

Abdel-Malek, T. (1974), “Foreign Ownership and Export Performance,” Journal of International

Business Studies”, Vol. 5, No. 2, pp.1-14.

Akhtar, S. (2005), “The Determinants of Capital Structure for Australian Multinational and Domestic

Corporations”, Australian Journal of Management, December 1, 2005.

Akhtar, S., and Barry, R. O. (2009), “Determinants of Capital Structure for Japanese Multinational and

Domestic Corporations”, International Review of Finance, Vol. 9, No.1-2, pp. 1-26.

Amsden, A.H. (2001), The Rise of the Rest: Challenges to the West from Late-Industrializing

Economies, (New York: Oxford University Press).

Anastassopoulos, G. (2003), “MNE Subsidiaries versus Domestic Enterprises: An Ananlysis of their

Ownership and Location-specific Advantages”, Applied Economics, Vol. 35, pp.1505-14.

Anastassopoulos, G. (2004), “Profitability Differences Between MNE Subsidiaries and Domestic

Firms: The Case of the Food Industry in Greece”, Agribusiness, Vol.20, No.1, pp.45-60.

Aydin, N., M. Sayim, A. Yalama (2007), “Foreign Ownership and Firm Performance”, International

Research Journal of Finance and Economics, Vol. 11, pp.103-11.

Barbosa, N., and H. Louri (2005), “Corporate Performance: Does Ownership Matter? A Comparison of

Foreign- and Domestic-Owned Firms in Greece and Portugal”, Review of Industrial Organization, Vol.

27, No.1, pp.73-102.

Basant, R. (1997), “Technology Strategies of Large Enterprises in Indian Industry: Some

Explorations”, World Development, Vol. 25, No. 10, pp.1683-1700.

Bellak, C. (2004a), “How Domestic and Foreign Firms Differ and Why Does it Matter?”, Journal of

Economic Survey, Vol. 18, No. 4, pp. 483-514.

Bellak, C. (2004b), “How Performance Gaps between Domestic Firms and Foreign Affiliates Matter

for Economic Policy”, Transnational Corporations, Vol. 13, No. 2, pp. 29-55.

49

Bellak, C. and Pfaffermayr, M. (2002), “Why Foreign-Owned Firms are Different: a Conceptual

Framework and Empirical Evidence for Austria”, in Jungnickel, R. (ed.), Foreign-owned Firms-Are

They Different?, pp. 13-57, (Houndsmill: Palgrave).

Bhaduri, S. and A. Ray (2004), “Exporting through Technological Capability: Econometric Evidence

from India’s Pharmaceutical and Electrical/Electronics Firms”, Oxford Development Studies, Vol. 32,

No.1, pp. 87-100.

Burgman, T. A. (1996), "An Empirical Examination of Multinational Corporate Capital Structure",

Journal of International Business Studies, Vol. 27, No. 3, pp. 553-70.

Casson, M. and R.D. Pearce (1987), "Multinational Enterprises in LDCs", in N. Gammell (ed.),

Surveys in Development Economics, pp. 90-132 , (Oxford: Basil Blackwel).

Castellani, D. and A. Zanfei (2006), Multinational Firms, Innovation and Productivity (Cheltenham:

Edward Elgar).

Caves, R. E. (1996), Multinational Enterprise and Economic Analysis, (Cambridge: Cambridge

University Press, second edition).

Chen, C. J. P., A. Cheng, N. He and J. Kim (1997), "An Investigation of the Relationship between

International Activities and Capital Structure", Journal of International Business Studies, Vol. 23, No.

3, pp. 563-77.

Chhibber, P. K and S. K. Majumdar (2005), "Property Rights and the Control of Strategy: Foreign

Ownership Rules and Domestic Firm Globalization in Indian Industry", Law & Policy, Vol. 27, No. 1,

pp.52-80.

Chhibber, P.K. and S. K. Majumdar (1999), "Foreign Ownership and Profitability: Property Rights,

Control, and the Performance of Firms in Indian Industry", Journal of Law and Economics, Vol. 42, pp.

209-38.

Chkir, I. E. and J. C. Cosset (2001), "Diversification Strategy and Capital Structure of Multinational

Corporations", Journal of Multinational Financial Management, Vol. 11, pp. 17-37.

Chuang, Y.C. and C. E. Lin (1999), "Foreign Direct Investment, R&D and Spillover Efficiency:

Evidence from Taiwan’s Manufacturing Firms", Journal of Development Studies, Vol. 35, No. 4,

pp.117-37.

50

Correa, P., M. Dayoub and M. Francisco (2007), “Identifying Supply-Side Constraints to Export

Performance in Ecuador: An Exercise with Investment Climate Survey Data”, World Bank Policy

Research Working Paper, No. 4179, March.

Dachs, B., B. Ebersberger, and H. Loof (2008), "The Innovative Performance of Foreign-owned

Enterprises in Small Open Economies", The Journal of Technology Transfer, Vol. 34, No. 4, pp. 393-

406.

Doukas, J.A. and C. Pantzalis (2003), "Geographic Diversification and Agency Costs of Debt of

Multinational Firms", Journal of Corporate Finance, Vol. 9, pp.59-92.

Douma, S., R. George, and R. Kabir (2006), Foreign and domestic Ownership, Business Groups and

Firm performance: Evidence from a Large Emerging Market", Strategic Management Journal, Vol. 27,

pp. 637–657.

Du, J. and S. Girma (2007). “Finance and Firm Export in China”, Kyklos, Vol. 60, No.1, pp.37-54.

Eiteman, D. K., A. I. Stonehill and H. M. Moffett (1998), Multinational Business Finance, (USA:

Addison Wesley, 8th Edition).

Fung, H., G. Y. Gao, J. Lu, and H. Mano (2008), “Impacts of Competitive Position on Export

Propensity and Intensity: An Empirical Study of Manufacturing Firms in China,” The Chinese

Economy, Vol. 41, No. 5, pp. 51-67.

GoI (2008), FDI in India Statistics, Department of Industrial Policy and Promotion, New Delhi.

Goldar, B., V.S. Renganathan and R. Banga (2004), "Ownership and Efficiency in Engineering Firms:

1990-91 to 1999-2000", Economic and Political Weekly, Review of Industry and Management, January

31, Vol. 39, No. 5, pp. 441-447.

Greenaway, D, A Guariglia, R Kneller (2007), "Financial Factor and Exporting Decisions", Journal of

International Economics, Vol. 73, No. 2, pp. 377-95.

Gujarati, D. N. (2004), Basic Econometrics, (New Delhi: Tata McGraw-Hill Publishing Company

Ltd.).

Haddad, M. and A. Harrison (1993), "Are There Positive Spillovers from Direct Foreign Investment?:

Evidence From Panel Data for Morocco", Journal of Development Economics, Vol. 42, pp. 51-74.

51

Helleiner, G.K. (1988), "Direct Foreign Investment and Manufacturing for Export in Developing

Counties: A Review of the Issues”, in S. Dell (ed.), Policies for Development, pp.123-153, (London:

Macmillan).

Hennart, J.-F. (1986), “What is Internalisation?”, Weltwirtschaftliches Archiv, Vol. 122, pp.791-806.

Homaifer, G., J. Zietz and O. Benkato (1998), "Determinants of Capital Structure in Multinational and

Domestic Corporations", Economia Internazionale, Vol. 51, No. 2, pp.189-210.

Hymer, S.H. (1976), The International Operation of National Firms: A Study of Direct Foreign

Investment, (Ph.D. Dissertation MIT, 1960), (Cambridge, Mass.: MIT Press).

Ito, Keiko, 2002. "Are Foreign Multinationals More Efficient? Plant Productivity in the Thai

Automobile Industry", Working Paper 2002-19, Kitakyushu: International Centre for the Study of East

Asian Development.

Jenkins, R. (1989), “Transnational Corporations, Competition and Monopoly", Review of Radical

Political Economy, Vol. 21, No. 4, pp.12-32.

Jenkins, R. (1990), “Comparing Foreign Subsidiaries and Local Firms in LDCs: Theoretical Issues and

Empirical Evidence”, The Journal of Development Studies, Vol. 26, No. 2, pp. 205-28.

Jensen, C. (2002), ‘Foreign Direct Investment, Industrial Restructuring and the Upgrading of Polish

Exports’, Applied Economics, Vol. 34, No.2, pp. 207–217.

Jensen, M.C. and Meckling, W.H., 1976, “Theory of the Firm: Managerial Behaviour, Agency Costs

and Ownership Structure”, Journal of Financial Economics, Vol. 3, pp. 305-360.

Jungnickel, R. (2002), Foreign-owned Firms: Are they Different?, (Houndmills: Palgrave-Macmillan).

Kathuria, V. (2001), "Foreign Firms, Technology Transfer and Knowledge Spillovers to Indian

Manufacturing Firms-A Stochastic Frontier Analysis", Applied Economics, Vol.33, No.5, pp. 625-42.

Kathuria, V. (2002), “Liberalisation, FDI and Productivity Spillovers-An Analysis of Indian

Manufacturing,” Oxford Economic Papers, Vol.54, No.4, 688-718.

52

Kathuria, V. (2008),“Impact of FDI Inflows on R&D Investment by Medium and High-tech Firms in

India in the Post Reform Period”, Transnational Corporations, Vol.17, No.2, 45-66.

Kneller R. and M. Pisu (2007), “Industrial linkages and Export Spillover from FDI”, World Economy,

Vol. 30, No.1, pp.105-34.

Kobrin, S. J. (2005), "The Determinants of Liberalization of FDI: Policy in Developing Countries: A

Cross-Sectional Analysis, 1992-2001, Transnational Corporations, Vol. 14, No.1, pp. 1-35.

Kokko, A., M. Zejan and R. Tansini (2001), Trade Regimes and Spillover Effects of FDI: Evidence

from Uruguay, Weltwirtschaftliches Archiv, Vol. 137, No.1, pp. 124-149.

Kokko, A., M. Zejan, R. Tansini (2001), “Trade Regimes and Spillover Effects from FDI: Evidence

from Uruguay”, Weltwirtschaftliches Archiv, Vol. 137, No. 1, pp. 15-32

Konings, J. (2001), “The Effects of Foreign Direct Investment on Domestic Firms: Evidence from Firm

level Panel Data in Eemerging economies”, Economics of Transition, Vol. 9: pp. 619–33.

Kuemmerle, W. (1999), "The Drivers of Foreign Direct Investment into Research and Development: an

Empirical Investigation", Journal of International Business Studies, Vol. 30, No.1, pp.1-25.

Kumar, N. (1990), Multinational Enterprises in India: Industrial Distribution, Characteristics and

Performance, (London and New York: Routledge).

Kumar, N. (1990). “Mobility Barriers and Profitability of Multinational and Local Enterprises in Indian

Manufacturing”, The Journal of Industrial Economics, Vol. 38, No. 4, pp. 449-63.

Kumar, N. (1991), “Mode of Rivalry and Comparative Behaviour of Multinationals and Local

Enterprises: The Case of Indian Manufacturing”, Journal of Development Economics, Vol. 35, pp. 381-

392.

Kumar, N. (2005), “Liberalisation, Foreign Direct Investment Flows and Development: Indian

Experience in the 1990s, Economic and Political weekly, April 2, pp.1459-69.

Kumar, N. and M. Saqib (1996), "Firm Size, Opportunities for Adaptation and in-House R&D Activity

in Developing Countries: The Case of Indian Manufacturing," Research Policy, Vol. 25, No. 5, pp. 712-

22.

Kumar, N. and N. S. Siddharthan (1997), Technology, Market Structure and Internationalization, (New

York and London: Routledge).

53

Kumari, A. (2007), "Liberalisation, TRIPS and Technological Behaviour of Firms in Indian

Pharmaceutical Industry", International Review of Business Research Papers, Vol.3, No.1, pp. 12 –22.

Lakshmanan, L., S. Chinngaihlian and R. Rajesh (2007), “Competitiveness of India’s Manufacturing

Sector: An Assessment of Related Issues”, Reserve Bank of India Occasional Papers, Vol. 28, No. 1,

pp. 33-77.

Lall, S. (1978), "Transnationals, Domestic Enterprises and Industrial Structure in LDCs: A Survey",

Oxford Economic Papers, Vol. 30, No. 2, pp. 217-48.

Lall, S. (1979), "Multinationals and Market Structure in an Open Developing Economy: The Case of

Malaysia", Weltwirtschaftliches Archiv, Vol. 115, pp. 325-48.

Lall, S. and R. Narula (2004), "Foreign Direct Investment and its Role in Economic Development: Do

We Need a New Agenda?", European Journal of Development Research, Vol. 16, No.3, pp. 447-64.

Lee, K. and C. Y. Kwok (1988), "Multinational Corporations vs. Domestic Corporations: International

Environmental Factors and Determinants of Capital Structure," Journal of International Business

Studies, Vol.19, pp.195-217.

Lennox, C. (1999), " Identifying Failing Companies: A Revaluation of the Logit, Probit and DA

Approaches", Journal of economics and Business, Vol. 51, pp. 347-64.

Markusen, J. R. (1995), “The Boundaries of Multinational Enterprises and the Theory of International

Trade,” The Journal of Economic Perspectives, Vol. 9, No. 2, pp. 169-189.

Menon, J., 1998, “Total Factor Productivity Growth in Foreign and Domestic Firms in Malaysian

Manufacturing,” Journal of Asian Economics, Vol. 9, pp. 251-80.

Mittoo, U.R. and Z. Zhang (2008), The Capital Structure of Multinational Corporations: Canadian

versus U.S. Evidence, Journal of Corporate Finance, Vol.14, No. 5, pp.706-20.

Modigliani, F. and M. H. Miller (1958), “The Cost of Capital, Corporation Finance and the Theory of

Investment,” American Economic Review, Vol. 53, pp. 261-97.

Mueller, D. C. (1986), Profits in the Long Run, (Cambridge: Cambridge University Press).

54

Myers, S. C. (1984), "The Capital Structure Puzzle," Journal of Finance, Vol.30, pp. 575-92.

Natke, P. A. (1987), "Foreign Ownership and Firm-level Import performance in Brazilian

Manufacturing Industries", Atlantic Economic Journal, Vol. 15, No. 4, pp. 39-48.

Nayyar, D. (1978), "Transnational Corporations and Manufactured Exports from Poor Countries",

Economic Journal, Vol. 8, No. 2, pp.59-84.

Newfarmer (1983), "Multinationals and Marketplace Magic in the 1980s" in C. P. Kindleberger and D.

Audretsch (eds.) The Multinational Corporation in the 1980s, (London: MIT Press).

Ngoc, P.M. and Ramstetter, E.D. (2004), “Foreign Multinationals and Local Firms in Vietnam’s

Economic Transition,” Asian Economic Journal, Vol. 18, No. 4, pp. 371-404.

OECD (2005), Internationalisation of R&D: Trends, Issues and Implications for S&T Policies: A

Review of Literature presented at the Forum on the Internationalisation of R&D, Brussels.

Oguchi, N., N. A. M. Amdzah, Z. Bakar, R. Z. Abidin and M. Shafii (2002), “Productivity of Foreign

and Domestic Firms in Malaysian Manufacturing Industry,” Asian Economic Journal, Vol. 16, pp. 215–

28.

Patibandala, M. and A. Sanyal (2005), “Foreign Investment and Productivity: A Study of Post-reform

Indian Industry”, Review of Applied Economics, Vol.1, No.1, pp. 1-14.

Penrose, E. (1959), The Theory of the Growth of the Firm, (New York: Wiley).

Pohar, M, M. Blas and S. Turk (2004), “Comparison of Logistic Regression and Linear Discriminant

Analysis: A Simulation Study,” Metodološki Zvezki, Vol. 1, No. 1, 143-161.

Pradhan, J. P. (2002b), "Liberalization, Firm Size and R&D performance: A Firm Level Study of Indian

Pharmaceutical Industry," Journal of Indian School of Political Economy, Vol. 14, No. 4, pp. 26-37

Ramstetter, E. D. (1994), “Comparisons of Japanese Multinationals and Other Firms in Thailand’s

Non-Oil Manufacturing Industries,” ASEAN Economic Bulletin, Vol.11, No.1, pp. 36-58.

55

Ramstetter, E. D. (1999a), "Comparisons of Foreign Multinationals and Local Firms in Asian

Manufacturing over Time," Asian Economic Journal, Vol. 13, No. 2, pp. 163–203.

Ramstetter, E. D. (1999b), "Trade Propensities and Foreign Ownership Shares in Indonesian

Manufacturing," Bulletin of Indonesian Economic Studies, 35, Vol. 2, pp. 43-66.

Ramstetter, E. D. (2002b), “Does Technology Differ in Local Plants and Foreign Multinationals in Thai

Manufacturing? Evidence from Translog Production Functions for 1996 and 1998”, Working Paper

Series 2002-04, Kitakyushu: The International Centre for the Study of East Asian Development.

Ramstetter, E. D. (2003), “Labor Productivity, Wages, Nationality, and Foreign Ownership Shares in

Thai manufacturing, 1996–2000, Journal of Asian Economics, Vol. 14, pp. 861-84.

Ramstetter, E. D. and A. Matsuoka (2001), "Recent Trends in Large Firms in Singapore and Thailand:

Did Foreign Multinationals Perform Differently than Local Firms through the Crisis?," in Toida M. and

D. Hiratsuka (eds), Projections for Asian Industrializing Region 2001, pp. 79–180, Institute of

Developing Economies, Tokyo.

Rasiah R. and A. Kumar (2008), "Foreign Ownership, Technological Intensities and Economic

Performance of Automotive Parts Firms in India", Asia Pacific Business Review Vol. 14, No. 1, pp. 85-

102.

Rasiah, R. (2004a), “Exports and Technological Capabilities: A Study of Foreign and Local Firms in

the Electronics Industry in Malaysia, the Philippines and Thailand”, European Journal of Development

Research, Vol. 16, No. 3, pp. 587–623.

Rasiah, R. (2004b), Foreign Firms, Technological Intensities and Economic Performance: Evidence

from Africa, Asia and Latin America, (Cheltenham: Edward Elgar).

Rasiah, R. and A. Malakolunthu (2009), "Technological Intensities and Economic Performance: A

Study of Foreign and Local Electronics Firms in Malaysia", Asia Pacific Business Review Vol. 15, No.

2, pp.181-97.

Rasiah, R. and Gachino (2005), “Are Foreign Firms More Productive and Export- and Technology-

intensive than Local Firms in Kenyan Manufacturing”, Oxford Development Studies, Vol. 33, No. 2, pp.

211-27.

Ray, A. S. and S. Bhaduri (2001), " R&D and Technological Learning in Indian Industry: Econometric

Estimation of Research Production Function," Oxford Development Studies, Vol. 29, No. 2, pp. 155-71.

56

Ray, P. K. and S. Rahman (2006), "The Propensity for Local Innovation and Inter-firm Linkages in

Transnational Corporations versus Local Enterprises in India," Transnational Corporations, Vol. 15,

No. 2, 71-98.

Ray, P. K. and S. Venaik (2001), “Foreign Firms and Local Linkages: A Study of Comparative

Behaviour of Multinational-affiliates and Local Enterprises in India”, Economic and Political Weekly,

Vol. 36, No. 46 & 47, pp. 4385-91.

Ray, P. K. and S. Venaik (2008), “Foreign Ownership and Subsidiaries Performance: Impact on

Research and Exports,” Economic and Political Weekly, Review of Industry and Management,

September 27, pp. M57-M65.

Ray, S. (2004), "MNEs, Strategic Alliances and Efficiency of Firms: Emerging Trends", Economic and

Political Weekly, Review of Industry and Management, Vol. 39, No. 5, pp. 434-40.

RBI (1985), Survey of Foreign Collaborations in Indian Industries, (Bombay: Reserve Bank of India).

Rugman, A. M. & Verbeke, A. (2001), "Subsidiary-specific Advantages in Multinational Enterprises,

Strategic Management Journal, Vol. 22, pp. 237-50.

Rugman, A. M. (1981), Inside the Multinationals: The Economics of Internal Market, (London: Croom

Helm).

Sasidharan, S. and A. Ramnathan (2007), "Foreign Direct Investment and Spillovers: Evidence from

Indian Manufacturing", International Journal of Trade and Global Markets, Vol. 1, No. 1, pp. 5-22.

Shapiro, A.C. (1996), Multinational Financial Management, 5th edition, (New Jersey. Prentice-Hall).

Siddharthan, N. S. (2009), "Determinants of R&D: Lessons from Literature," Working Paper, Forum

for Global Knowledge Sharing and Madras School of Economics, pp.1-34.

Siddharthan, N.S. and K. L. Krishna (1994), “Determinants of Technology Imports: Evidence for

Indian Firms”, Working Paper Series No. E/161/94, Institute of Economic Growth, Delhi.

Siddharthan, N.S. and S. Nollen (2004), "MNE Affiliation, Firm Size and Exports Revisited: A Study

of Information Technology Firms in India", The Journal of Development Studies, Vol. 40, No.6, pp.

146-68.

Singh, K. and J. E. Hodder (2000) “Multinational Capital Structure and Financial Flexibility”, Journal

of International Money & Finance, Vol. 19, No.6, pp. 853-85.

57

Sinha, R. (1993), "Foreign Participation and Technical Efficiency", Applied Economics, Vo. 25, pp.583-

88.

Sjöholm, F. (1999a), "Technology Gap, Competition and Spillovers from Direct Foreign Investment:

Evidence from Establishment Data", Journal of Development Studies, Vol. 36, No.1, pp. 53-73.

Smeets, R. (2008), “Collecting the Pieces of the FDI Knowledge Spillovers Puzzle”, The World Bank

Research Observer, Vol. 23, No. 2, pp.107-38.

Sun, S. (2009), "How Does FDI Affect Domestic Firms’ Exports? Industrial Evidence", World

Economy, Vol. 32, No. 8, pp.1203-22.

Takii, S. (2004), “Productivity Differentials between Local and Foreign Plants in Indonesian

Manufacturing, 1995”, World Development, Vol. 32, No. 11, pp. 1957- 69.

Takii, S. and E. D. Ramstetter (2003), Employment, Production, Labor Productivity, and Foreign

Multinationals in Indonesian Manufacturing, 1975–2000, Working Paper No. 2003-25, International

Centre for the Study of East Asian Development, Kitakyushu.

Tambunlertchai, S. and E. D. Ramstetter (1991), “Foreign Firms in Promoted Industries and Structural

Change in Thailand,” in E. D. Ramstetter (ed), Direct Foreign Investment in Asia’s Developing

Economies and Structural Change in the Asia-Pacific Region, pp. 65-104, (Boulder:Westview Press).

UNCTAD (2005), World Investment Report 2005: Transnational Corporations and the

Internationalisation of R&D, (New York and Geneva: United Nations).

Wang, C. & L. Yu (2007), “Do Spillover Benefits Grow with Rising Foreign Direct Investment? An

Empirical Examination of the Case of China”, Applied Economics, Vol. 39, pp. 397-405.

Wignaraja, G. (2008), “Foreign Ownership, Technological Capabilities and Clothing Exports in Sri

Lanka", Journal of Asian Economics, Vol.19, pp. 29-39.

Wiwattanakantang, Y. (2001), “Controlling Shareholders and Corporate Value: Evidence from

Thailand”, Pacific-Basin Finance Journal, Vol. 9, pp. 323-62.

58